Merge pull request #620 from Routstr/fix/fee-payout-crash-guard

fix: guard fee payouts against crash double-payments

.env.example

@@ -2,6 +2,9 @@
 UPSTREAM_BASE_URL=https://api.openai.com/v1
 UPSTREAM_API_KEY=your-upstream-api-key
 
+# Tinfoil (confidential inference enclaves, EHBP)
+# TINFOIL_API_KEY=your-tinfoil-api-key
+
 # ADMIN_PASSWORD=secure-admin-password
 
 # Database
@@ -14,6 +17,7 @@ UPSTREAM_API_KEY=your-upstream-api-key
 # HTTP_URL=https://api.mynode.com
 # ONION_URL=http://mynode.onion (auto fetched from compose)
 # RELAYS="wss://relay.damus.io,wss://relay.nostr.band,wss://eden.nostr.land,wss://relay.routstr.com"
+# ENABLE_ANALYTICS_SHARING=true
 # CASHU_MINTS="https://mint.minibits.cash/Bitcoin,https://mint.cubabitcoin.org,https://ecashmint.otrta.me"
 # RECEIVE_LN_ADDRESS=
 

.github/workflows/container.yml

@@ -14,6 +14,14 @@ jobs:
     steps:
       - name: Checkout code
         uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+
+      - name: Resolve git metadata
+        id: gitmeta
+        run: |
+          echo "sha=$(git rev-parse --short=7 HEAD)" >> "$GITHUB_OUTPUT"
+          echo "tag=$(git describe --tags --exact-match HEAD 2>/dev/null || true)" >> "$GITHUB_OUTPUT"
 
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v2
@@ -29,7 +37,11 @@ jobs:
         uses: docker/build-push-action@v4
         with:
           context: .
+          file: Dockerfile.full
           push: true
+          build-args: |
+            GIT_COMMIT=${{ steps.gitmeta.outputs.sha }}
+            GIT_TAG=${{ steps.gitmeta.outputs.tag }}
           tags: |
             ghcr.io/routstr/proxy:latest
             ghcr.io/routstr/core:latest

.github/workflows/test.yml

@@ -67,7 +67,7 @@ jobs:
       - name: Setup Node.js
         uses: actions/setup-node@v4
         with:
-          node-version: "18"
+          node-version: "20"
           cache: "pnpm"
           cache-dependency-path: ui/pnpm-lock.yaml
 

.gitignore

@@ -16,6 +16,9 @@ dist/
 *.db-shm
 *.db-wal
 .*wallet.sqlite3
+.wallet/
+AGENTS.md
+TEST_SUITE_OVERVIEW.md
 *models.json
 .cashu
 .relay
@@ -38,3 +41,4 @@ proof_backups
 
 *.todo
 ui_out
+.worktrees

Dockerfile

@@ -1,26 +1,29 @@
-FROM ghcr.io/astral-sh/uv:python3.11-alpine
+FROM ghcr.io/astral-sh/uv:python3.11-bookworm-slim
 
-# Install system dependencies required for secp256k1
-RUN apk add --no-cache \
-    pkgconf \
-    build-base \
-    automake \
-    autoconf \
-    libtool \
-    m4 \
-    perl
-RUN apk add git
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+        git \
+        build-essential \
+        pkg-config \
+        libsecp256k1-dev \
+        autoconf \
+        automake \
+        libtool \
+    && rm -rf /var/lib/apt/lists/*
 
 COPY uv.lock pyproject.toml ./
 RUN mkdir -p /routstr
 
-RUN uv add git+https://github.com/saschanaz/secp256k1-py.git#branch=upgrade060
-# RUN uv sync
+RUN uv sync --frozen --no-dev --no-install-project
 
 WORKDIR /app
 
 COPY . .
 
+ARG GIT_COMMIT=""
+ARG GIT_TAG=""
+ENV GIT_COMMIT=${GIT_COMMIT}
+ENV GIT_TAG=${GIT_TAG}
 ENV PORT=8000
 ENV PYTHONUNBUFFERED=1
 

Dockerfile.full

@@ -4,10 +4,10 @@ FROM node:23-alpine AS ui-builder
 WORKDIR /app/ui
 
 # Install pnpm
-RUN corepack enable pnpm && corepack prepare pnpm@latest --activate
+RUN corepack enable pnpm && corepack prepare pnpm@10.15.0 --activate
 
 # Copy UI source
-COPY ui/package.json ui/pnpm-lock.yaml* ./
+COPY ui/package.json ui/pnpm-lock.yaml* ui/pnpm-workspace.yaml* ./
 RUN pnpm install --frozen-lockfile
 
 COPY ui/ ./
@@ -16,35 +16,38 @@ ENV NEXT_TELEMETRY_DISABLED=1
 RUN pnpm run build
 
 # Stage 2: Build the Routstr Node
-FROM ghcr.io/astral-sh/uv:python3.11-alpine AS runner
+FROM ghcr.io/astral-sh/uv:python3.11-bookworm-slim AS runner
 
-# Install system dependencies
-RUN apk add --no-cache \
-    pkgconf \
-    build-base \
-    automake \
-    autoconf \
-    libtool \
-    m4 \
-    perl \
-    git
+RUN apt-get update \
+    && apt-get install -y --no-install-recommends \
+        git \
+        build-essential \
+        pkg-config \
+        libsecp256k1-dev \
+        autoconf \
+        automake \
+        libtool \
+    && rm -rf /var/lib/apt/lists/*
+
+COPY uv.lock pyproject.toml ./
+
+RUN uv sync --no-dev --no-install-project
 
 WORKDIR /app
 
-# Copy the rest of the application (required for uv sync to find the package)
 COPY . .
 
-# Install dependencies including the specific secp256k1 branch
-RUN uv add git+https://github.com/saschanaz/secp256k1-py.git#branch=upgrade060
-RUN uv sync --no-dev
-
 # Copy the built UI from the ui-builder stage
 COPY --from=ui-builder /app/ui/out ./ui_out
 
+ARG GIT_COMMIT=""
+ARG GIT_TAG=""
+ENV GIT_COMMIT=${GIT_COMMIT}
+ENV GIT_TAG=${GIT_TAG}
 ENV PORT=8000
 ENV PYTHONUNBUFFERED=1
 
 EXPOSE 8000
 
 # Run the application
-CMD ["/app/.venv/bin/fastapi", "run", "routstr", "--host", "0.0.0.0"]
+CMD ["/.venv/bin/fastapi", "run", "routstr", "--host", "0.0.0.0"]

Makefile

@@ -98,7 +98,7 @@ docker-down:
 lint:
 	@echo "🔍 Running linting checks..."
 	$(RUFF) check .
-	$(MYPY) routstr/ --ignore-missing-imports
+	$(MYPY) .
 
 format:
 	@echo "✨ Formatting code..."
@@ -107,7 +107,7 @@ format:
 
 type-check:
 	@echo "🔎 Running type checks..."
-	$(MYPY) routstr/ --ignore-missing-imports
+	$(MYPY) .
 
 # Development setup
 dev-setup:
@@ -234,7 +234,7 @@ ci-test:
 ci-lint:
 	@echo "🤖 Running CI linting..."
 	$(RUFF) check . --exit-non-zero-on-fix
-	$(MYPY) routstr/ --ignore-missing-imports --no-error-summary
+	$(MYPY) . --no-error-summary
 
 # Debug helpers
 test-debug:

README.md

@@ -51,14 +51,26 @@ curl https://api.routstr.com/v1/chat/completions \
 
 ## Quick Start (Docker)
 
-If you are a node runner, start a Routstr Core instance and configure upstream access in the dashboard.
+If you are a node runner, start a Routstr Core instance using Docker Compose:
 
-```bash
-docker run -d \
-  --name routstr-proxy \
-  -p 8000:8000 \
-  ghcr.io/routstr/proxy:latest
-```
+1. **Prepare your `.env`**:
+   ```bash
+   ADMIN_PASSWORD=mysecretpassword
+   NAME="My AI Node"
+   DESCRIPTION="Fast access to models"
+   NSEC=yournsec
+   RECEIVE_LN_ADDRESS=yourname@wallet.com
+   ```
+
+2. **Start the services**:
+   ```bash
+   docker compose up -d
+   ```
+
+3. **Configure**:
+   Open [http://localhost:8000/admin/](http://localhost:8000/admin/) to connect your AI providers and set pricing.
+
+For full instructions, see the **[Provider Quick Start Guide](https://docs.routstr.com/provider/quickstart/)**.
 
 ## Development
 

compose.yml

@@ -8,20 +8,22 @@ services:
       args:
         # NEXT_PUBLIC_API_URL: ${NEXT_PUBLIC_API_URL:-http://127.0.0.1:8000}
         NEXT_PUBLIC_ADMIN_API_KEY: ${NEXT_PUBLIC_ADMIN_API_KEY:-}
+    user: root
     volumes:
-      - ./ui_out:/output
+      - ./ui_out:/output:z
     command:
       ["sh", "-c", "mkdir -p /output && cp -r /app/built/. /output/ && echo 'UI build copied to mounted volume' && ls -la /output/ && echo 'UI built and ready' && tail -f /dev/null"]
+    restart: unless-stopped
 
   routstr:
     build: .
     depends_on:
       - ui
     volumes:
-      - .:/app
-      - ./logs:/app/logs
+      - .:/app:z
+      - ./logs:/app/logs:z
       - tor-data:/var/lib/tor:ro
-      - ./ui_out:/app/ui_out:ro
+      - ./ui_out:/app/ui_out:ro,z
     env_file:
       - .env
     environment:
@@ -30,6 +32,7 @@ services:
       - 8000:8000
     extra_hosts: # Needed to access locally running models
       - "host.docker.internal:host-gateway"
+    restart: unless-stopped
 
   tor:
     image: ghcr.io/hundehausen/tor-hidden-service:latest
@@ -40,6 +43,7 @@ services:
       - HS_ROUTER=routstr:8000:80
     depends_on:
       - routstr
+    restart: unless-stopped
 
 volumes:
   tor-data:

docs/api/endpoints.md

@@ -396,6 +396,8 @@ Authorization: Bearer sk-...
 }
 ```
 
+`balance` is the spendable balance used by request admission.
+
 ### Check Balance
 
 Get current wallet balance.

docs/api/errors.md

@@ -113,42 +113,109 @@ All errors follow a consistent JSON structure:
 **Status:** 402  
 **Resolution:** Top up API key balance
 
-#### Invalid Token
+### Cashu Token Redemption Errors
+
+These errors are returned when a Cashu token you pay with cannot be redeemed.
+They apply to every endpoint that accepts a token:
+
+- **Per-request payment** via the `X-Cashu` header (chat completions + Responses API).
+- **API key top-up** via `POST /v1/wallet/topup`.
+- **Minting an API key** from a token sent in `Authorization: Bearer <cashu-token>`.
+
+All three share one classifier, so the same failure yields the same HTTP status
+and sanitized message everywhere. Structured error envelopes (`X-Cashu` and
+`Authorization: Bearer <cashu-token>`) also expose the same `type` and `code` —
+branch on `type` (or `code` for finer granularity). `POST /v1/wallet/topup`
+keeps its existing plain-string `detail` envelope, so branch on status there.
+
+| `type` | Status | `code` | Retryable | Meaning |
+|--------|--------|--------|-----------|---------|
+| `token_already_spent` | 400 | `cashu_token_already_spent` | No | The token was already redeemed. |
+| `invalid_token` | 400 | `invalid_cashu_token` | No | The token is malformed or cannot be decoded. |
+| `mint_error` | 422 | `cashu_token_swap_fees_exceed_amount` | No | Token value is too small to cover the mint's swap/melt fees. |
+| `mint_error` | 422 | `cashu_foreign_mint_swap_failed` | No | Swapping the token from a foreign mint to the primary mint failed. |
+| `mint_unreachable` | 503 | `cashu_mint_unreachable` | **Yes** | The mint could not be reached (DNS failure, refused/reset connection, timeout). The token is fine — retry once the mint recovers. |
+| `cashu_error` | 400 | `cashu_token_redemption_failed` | No | The token could not be redeemed for another expected reason. |
+| `cashu_error` | 400 | `cashu_token_zero_value` | No | The token redeemed to zero (empty/dust token, or value fully consumed by fees). |
+| `token_consumed` | 500 | `cashu_token_consumed` | No | The token was **spent** (melted/redeemed) but crediting it then failed. Do not retry — the token is gone; contact support to reconcile. |
+| `api_error` | 500 | `internal_error` | Maybe | Unexpected server-side fault during redemption. |
+
+!!! important "Retry only `mint_unreachable`"
+    Only `mint_unreachable` (503) means the same token will work again later —
+    everything else is a permanent property of the token and must not be
+    blindly retried. Use exponential backoff for the 503. In particular, a
+    `token_consumed` 500 means the mint already spent the token, so a retry
+    would fail as `token_already_spent`.
+
+#### Mint Unreachable (retryable)
 
 ```json
 {
   "error": {
-    "type": "payment_error",
-    "message": "Invalid Cashu token",
-    "code": "invalid_token",
-    "details": {
-      "reason": "Token already spent"
-    }
+    "type": "mint_unreachable",
+    "message": "Cashu mint is unreachable",
+    "code": "cashu_mint_unreachable"
   }
 }
 ```
 
-**Status:** 400  
-**Resolution:** Use a valid, unspent token
+**Status:** 503
 
-#### Mint Unavailable
+**Resolution:** The token is valid — the mint is temporarily down. Retry with
+backoff, or pay with a token from a different mint.
+
+#### Token Already Spent
 
 ```json
 {
   "error": {
-    "type": "payment_error",
-    "message": "Cannot connect to Cashu mint",
-    "code": "mint_unavailable",
-    "details": {
-      "mint_url": "https://mint.example.com",
-      "retry_after": 60
-    }
+    "type": "token_already_spent",
+    "message": "Cashu token already spent",
+    "code": "cashu_token_already_spent"
   }
 }
 ```
 
-**Status:** 503  
-**Resolution:** Try again later or use different mint
+**Status:** 400
+
+**Resolution:** Use a fresh, unspent token. Do not retry with the same token.
+
+#### Response envelope differs by endpoint
+
+The `error` object above is identical everywhere, but the surrounding envelope
+depends on how you paid:
+
+- **`X-Cashu` header payments** (chat + Responses API) return the object at the
+  top level, alongside a `request_id`:
+
+  ```json
+  {
+    "error": { "type": "mint_unreachable", "message": "Cashu mint is unreachable", "code": "cashu_mint_unreachable" },
+    "request_id": "req-abc123"
+  }
+  ```
+
+  The original token is echoed back in the `X-Cashu` **response header only when
+  it is still spendable** (e.g. `mint_unreachable`, `invalid_cashu_token`, fee
+  errors) so you can recover/retry it. It is **not** echoed for spent/consumed
+  tokens (`cashu_token_already_spent`, `cashu_token_consumed`,
+  `cashu_token_zero_value`, `internal_error`) — retrying those can never succeed.
+
+- **`Authorization: Bearer <cashu-token>`** (API key minting) wraps it in
+  FastAPI's `detail` field:
+
+  ```json
+  { "detail": { "error": { "type": "mint_unreachable", "message": "Cashu mint is unreachable", "code": "cashu_mint_unreachable" } } }
+  ```
+
+- **`POST /v1/wallet/topup`** returns a plain string message under `detail` —
+  it carries the shared HTTP **status** and **message** (e.g. `503` for an
+  unreachable mint) but not the structured `type`/`code`, so branch on the
+  status code here:
+
+  ```json
+  { "detail": "Cashu mint is unreachable" }
+  ```
 
 ### Validation Errors
 
@@ -347,7 +414,7 @@ class ErrorHandler:
         'rate_limit',
         'upstream_timeout',
         'model_overloaded',
-        'mint_unavailable'
+        'cashu_mint_unreachable'
     }
     
     # Errors requiring user action

docs/api/overview.md

@@ -155,7 +155,7 @@ The response includes your change in the same header:
 X-Cashu: cashuA7k2mNp4...
 ```
 
-This is fully stateless—no session, no `/v1/balance/refund` call needed. However, **streaming does not work with `X-Cashu`** because the refund can only be calculated after the full response is generated.
+This is fully stateless—no session, no `/v1/balance/refund` call needed. However, **streaming does not work with `X-Cashu`** because the refund can only be calculated after the full response is generated. If you lose the `X-Cashu` response header before claiming your change, you can reclaim the refund via `POST /v1/wallet/refund` by supplying the original payment token in the `x-cashu` header.
 
 ## Response Headers
 

docs/client/payments.md

@@ -53,9 +53,11 @@ If your balance runs low, you don't need a new key. You can top up the existing
 
 ### Via Lightning
 
-`POST /lightning/invoice` with `{"amount_sats": 1000, "purpose": "topup", "api_key": "sk-..."}`.
+`POST /lightning/invoice` with `Authorization: Bearer sk-...` header and body `{"amount_sats": 1000, "purpose": "topup"}`.
 *Once paid, the funds are added to your existing key.*
 
+> Legacy: the endpoint is also exposed at `/v1/balance/lightning/invoice`, and accepts an `api_key` field in the body as a fallback for older clients. New integrations should use the RIP-08 path with the `Authorization` header.
+
 ### Via Cashu
 
 `POST /v1/balance/topup` with `{"cashu_token": "..."}` and `Authorization: Bearer sk-...`.

docs/ehbp-proxy-support.md

@@ -0,0 +1,150 @@
+# EHBP Proxy Support for Tinfoil Models
+
+## Problem
+
+The SDK's `SecureClient.fetch` encrypts request bodies with HPKE (EHBP protocol)
+and sends them to the Routstr provider. The Routstr proxy had no EHBP handling:
+
+1. It tried to `json.loads()` the binary HPKE-sealed body → failed with a 400
+2. The upstream PPQ.AI public endpoint (`/v1/chat/completions`) doesn't speak
+   EHBP, so the response had no `Ehbp-Response-Nonce` header
+3. `SecureClient` threw `Missing Ehbp-Response-Nonce header` because it expects
+   every response from an EHBP-configured `baseURL` to carry that header
+
+## Root cause
+
+PPQ.AI exposes EHBP-aware inference at `/private/`, separate from the public
+`/v1/` endpoint. The Routstr proxy was forwarding to `/v1/` (the public
+endpoint) instead of `/private/` (the enclave endpoint). The public endpoint
+can't decrypt the body, returns a normal HTTP response, and the SDK can't
+decrypt it because there's no nonce header.
+
+The PPQ private-mode proxy (`ppq-private-mode-proxy/lib/proxy.ts`) shows the
+correct pattern: `SecureClient` talks to `api.ppq.ai/private/v1/chat/completions`,
+which decrypts inside the attested enclave and returns an EHBP-encrypted
+response with the `Ehbp-Response-Nonce` header.
+
+## What was changed
+
+### `routstr/proxy.py`
+
+Detects EHBP requests by checking for the `Ehbp-Encapsulated-Key` header (set
+by the EHBP transport on every encrypted request). For EHBP requests:
+
+- Skips JSON body parsing (the body is binary ciphertext, not JSON)
+- Reads the model ID from the `X-Routstr-Model` header (set by the SDK) instead
+  of from `body.model`
+- Routes through new `forward_ehbp_request` (bearer auth) and
+  `forward_ehbp_x_cashu_request` (x-cashu auth) methods
+- Skips reactive 400 param correction (can't parse encrypted response body)
+- Still charges the user via `pay_for_request` (uses `max_cost_for_model` from
+  the model registry, not the body)
+
+### `routstr/upstream/base.py`
+
+Keeps EHBP as an explicit opt-in provider capability instead of making every
+upstream provider appear EHBP-capable:
+
+- `supports_ehbp = False` by default
+- `get_ehbp_forwarding_target(path, model_obj)` raises `NotImplementedError`
+  unless a provider opts in and returns a provider-specific EHBP target
+
+The actual EHBP forwarding logic does **not** live in `base.py`.
+
+### `routstr/upstream/ehbp.py`
+
+Contains the shared opaque EHBP transport and billing helpers:
+
+- `EHBPForwardingTarget` — provider-specific target URL plus extra headers
+- `forward_ehbp_request()` — forwards the encrypted body, captures Tinfoil
+  usage from a response header or streaming HTTP trailer, and finalizes bearer
+  billing at actual cost (falling back to max cost when usage is unavailable)
+- `forward_ehbp_x_cashu_request()` — redeems the Cashu token, refunds the full
+  token on upstream failure, and refunds the difference between the redeemed
+  amount and actual cost (or max cost when usage is unavailable)
+
+### Provider support
+
+EHBP is currently enabled only for `TinfoilUpstreamProvider`. It forwards to
+Tinfoil's attested enclave and requests `X-Tinfoil-Usage-Metrics` for billing.
+PPQ.AI retains its private-target implementation, but `supports_ehbp = False`
+until it has a provider-specific trusted usage/model-binding strategy.
+
+## Why it's done this way
+
+The proxy is a **blind relay** for EHBP requests. It cannot decrypt the body
+(only the attested enclave can), so it must:
+
+1. Get the model ID from a header, not the body
+2. Forward the raw bytes without parsing or transformation
+3. Stream the response back without SSE/cost parsing
+4. Pass through EHBP protocol headers (`Ehbp-Encapsulated-Key` on request,
+   `Ehbp-Response-Nonce` on response)
+
+Cost tracking happens at the proxy level. Routstr reserves or redeems up to
+`max_cost_for_model`, then Tinfoil's out-of-band usage header/trailer allows it
+to finalize at actual token cost. If trusted usage is missing or invalid, the
+proxy safely falls back to max-cost billing.
+
+## End-to-end flow
+
+```
+SDK                                  Routstr Proxy                    PPQ.AI /private/
+  │                                       │                                │
+  │── X-Routstr-Model: tinfoil-kimi-k2-6 ─│                                │
+  │── Ehbp-Encapsulated-Key: <hex> ───────│                                │
+  │── Authorization: Bearer <cashu> ──────│                                │
+  │── body = HPKE-encrypted(kimi-k2-6) ───│                                │
+  │                                       │                                │
+  │                                  detects Ehbp-Encapsulated-Key         │
+  │                                  reads model from X-Routstr-Model      │
+  │                                  does billing/routing                  │
+  │                                       │                                │
+  │                                  adds X-Private-Model: private/kimi-k2-6
+  │                                  forwards raw body to /private/v1/...  │
+  │                                       │──────────────────────────────▶│
+  │                                       │                          enclave decrypts
+  │                                       │                          runs inference
+  │                                       │◀── Ehbp-Response-Nonce ──────│
+  │                                       │◀── encrypted response ────────│
+  │                                       │                                │
+  │                                  streams response back untouched       │
+  │◀── encrypted response ────────────────│                                │
+  │                                        │                                │
+  SecureClient reads nonce, decrypts       │                                │
+  SDK SSE processing sees plaintext        │                                │
+```
+
+## Model ID mapping
+
+Three parties see three different model IDs:
+
+| Party | Header/Body | Value | Source |
+|---|---|---|---|
+| Routstr proxy | `X-Routstr-Model` header | `tinfoil-kimi-k2-6` | SDK sends full caller-facing id |
+| Tinfoil usage metrics | `model` field | `kimi-k2-6` | Enclave reports the model actually served |
+| Tinfoil enclave | `body.model` (encrypted) | `kimi-k2-6` | SDK strips `tinfoil-` prefix before encryption |
+
+## Implementation status
+
+A dedicated `TinfoilUpstreamProvider` (`routstr/upstream/tinfoil.py`) now
+implements the direct blind-upstream pattern described above. The shared EHBP
+helpers in `routstr/upstream/ehbp.py` were extended to:
+
+- Request usage metrics via `X-Tinfoil-Request-Usage-Metrics: true`.
+- Parse `X-Tinfoil-Usage-Metrics` from the response header (non-streaming) or
+  HTTP trailer (streaming).
+- Override the forwarding URL with a validated `X-Tinfoil-Enclave-Url` when the
+  SDK sends it.
+- Finalize bearer billing with the dedicated EHBP actual-cost finalizer.
+- Compute X-Cashu refunds from actual cost instead of max cost.
+
+See `docs/tinfoil-direct-integration.md` for the full implementation notes.
+
+## Verification status
+
+Unit coverage includes usage parsing, target validation, HTTP trailer capture,
+response-size limits, and bearer payment finalization. End-to-end requests have
+verified both non-streaming usage headers and streaming usage trailers against
+Tinfoil. SDK behavior on proxy-generated non-2xx responses without an
+`Ehbp-Response-Nonce` still merits explicit end-to-end coverage.

docs/provider/configuration.md

@@ -78,9 +78,15 @@ Which mints to accept payments from:
 
 Automatic profit withdrawal:
 
-| Setting               | Description                     |
-| --------------------- | ------------------------------- |
-| **Lightning Address** | Your LN address for withdrawals |
+| Setting                              | Description                                                                                                | Default |
+| ------------------------------------ | ---------------------------------------------------------------------------------------------------------- | ------- |
+| **Lightning Address**                | Your LN address for withdrawals                                                                            | —       |
+| **Minimum Payout (sat)**             | Min available balance (in sats) before profit is paid out. Applies to both `sat` and `msat` mints (auto-converted). | `210`   |
+| **Payout Interval (seconds)**        | How often the payout loop wakes up and checks balances                                                     | `900`   |
+
+All payout amounts must be positive. Set the minimums above your wallet's
+minimum-invoice constraint (typically 1 sat) and high enough to amortise
+routing fees.
 
 ### Security
 
@@ -97,6 +103,7 @@ Announce your node on the network:
 | **Npub**   | Your Nostr public key                |
 | **Nsec**   | Your Nostr private key (for signing) |
 | **Relays** | Relays to publish announcements      |
+| **Share Analytics** | Publish aggregate usage stats to Nostr |
 
 See [Discovery](discovery.md) for details.
 
@@ -122,8 +129,11 @@ Use environment variables for:
 | `DESCRIPTION`        | Node description                  | `A Routstr Node`                     |
 | `NPUB`               | Nostr public key (bech32)         | —                                    |
 | `NSEC`               | Nostr private key                 | —                                    |
+| `ENABLE_ANALYTICS_SHARING` | Enable usage analytics sharing to Nostr | `true`                         |
 | `CASHU_MINTS`        | Comma-separated mint URLs         | `https://mint.minibits.cash/Bitcoin` |
 | `RECEIVE_LN_ADDRESS` | Lightning address for withdrawals | —                                    |
+| `MIN_PAYOUT_SAT`     | Min payout balance in sats (applies to all mints) | `210`                |
+| `PAYOUT_INTERVAL_SECONDS` | Payout loop interval (seconds) | `900`                            |
 | `TOR_PROXY_URL`      | SOCKS5 proxy for Tor              | `socks5://127.0.0.1:9050`            |
 | `CORS_ORIGINS`       | Allowed CORS origins              | `*`                                  |
 | `RELAYS`             | Nostr relays (comma-separated)    | (default set)                        |

docs/provider/dashboard.md

@@ -152,6 +152,7 @@ Manage which mints you accept payments from:
 |-------|-------------|
 | **Nsec** | Private key for signing announcements |
 | **Relays** | Where to publish your node advertisement |
+| **Share Analytics** | Toggle publishing aggregate usage stats to Nostr |
 
 ### Security
 

docs/provider/deployment.md

@@ -2,20 +2,62 @@
 
 Production deployment guide for Routstr Provider nodes.
 
+## All-in-One Docker Image (Preferred)
+
+The easiest way to deploy Routstr is using the all-in-one Docker image from Docker Hub, which includes both the FastAPI backend and the Next.js admin dashboard in a single container.
+
+### Quick Start
+
+```bash
+docker run -d \
+  --name routstr \
+  -p 8000:8000 \
+  -v routstr-data:/app/data \
+  -e DATABASE_URL="sqlite:////app/data/routstr.db" \
+  9qeklajc/routstr:latest
+```
+
+Access your node:
+- **API & Admin Dashboard**: http://localhost:8000
+
+### Docker Compose Setup
+
+Create `docker-compose.yml`:
+
+```yaml
+version: '3.8'
+
+services:
+  routstr:
+    image: 9qeklajc/routstr:latest
+    container_name: routstr
+    restart: unless-stopped
+    ports:
+      - "8000:8000"
+    volumes:
+      - routstr-data:/app/data
+    environment:
+      DATABASE_URL: "sqlite:////app/data/routstr.db"
+      ADMIN_KEY: "your-secure-admin-key"
+      LOG_LEVEL: "info"
+
+volumes:
+  routstr-data:
+```
+
+Start it:
+
+```bash
+docker compose up -d
+```
+
+---
+
 ## Docker Compose (Recommended)
 
 For production, use Docker Compose with persistent storage and optional Tor support.
 
-### Unified Setup (All-in-one)
-To build and run the node with the UI integrated in a single container using the multi-stage build:
-
-```bash
-docker build -f Dockerfile.full -t routstr-full .
-docker run -d -p 8000:8000 --env-file .env routstr-full
-```
-
-### Advanced Setup (Separated UI & Node)
-Use the included `compose.yml` for a more flexible setup that separates the UI build process from the node execution. This is useful for development or when you want to manage Tor as a separate service.
+Use the included `compose.yml` for a flexible setup that handles both the UI and the node execution. This is useful for development or when you want to manage Tor as a separate service.
 
 ```bash
 docker compose up -d
@@ -184,20 +226,16 @@ docker compose up -d
 
 ## Building from Source
 
-### Unified Image (UI + Node)
-The easiest way to build everything from source into a single production-ready image:
+### Using Docker Compose
+The easiest way to build everything from source:
 
 ```bash
-docker build -f Dockerfile.full -t routstr-full .
+docker compose build
 ```
 
 ### Individual Components
-If you prefer building them separately or using Docker Compose:
+If you prefer building the node only (requires manual UI build first):
 
 ```bash
-# Build using compose
-docker compose build
-
-# Or build the node only (requires manual UI build first)
 docker build -t routstr-node .
 ```

docs/provider/quickstart.md

@@ -35,6 +35,7 @@ ADMIN_PASSWORD=mysecretpassword
 # Node Identity
 NAME="My AI Node"
 DESCRIPTION="Fast access to models"
+NSEC=yournsec
 
 # Lightning Payouts
 RECEIVE_LN_ADDRESS=yourname@wallet.com
@@ -43,32 +44,10 @@ RECEIVE_LN_ADDRESS=yourname@wallet.com
 
 ## 2. Start the Node
 
-You can run the pre-built image directly:
+The recommended way to run Routstr is using Docker Compose, which handles the node, the UI, and optional services like Tor.
 
 ```bash
-docker run -d \
-  --name routstr \
-  -p 8000:8000 \
-  --env-file .env \
-  -v routstr-data:/app/data \
-  ghcr.io/routstr/proxy:latest
-```
-
-*Note: The pre-built image does not contain the UI. For the all-in-one experience with the Admin Dashboard, use the Build from Source instructions below.*
-
-### Build from Source (Recommended)
-
-If you want to build the node and UI yourself from source, use the unified Dockerfile:
-
-```bash
-git clone https://github.com/routstr/routstr-core.git
-cd routstr-core
-# Edit your .env with ADMIN_PASSWORD and API keys
-cp .env.example .env
-nano .env
-
-docker build -f Dockerfile.full -t routstr-local .
-docker run -d -p 8000:8000 --env-file .env --name routstr routstr-local
+docker compose up -d
 ```
 
 Verify it's running:
@@ -77,6 +56,15 @@ Verify it's running:
 curl http://localhost:8000/v1/info
 ```
 
+### Build from Source (Optional)
+
+If you've cloned the repository and want to build the images yourself:
+
+```bash
+docker compose build
+docker compose up -d
+```
+
 ---
 
 ## 3. Configure via Dashboard

docs/tinfoil-direct-integration.md

@@ -0,0 +1,493 @@
+# Tinfoil / PPQ Private-Mode Integration Notes
+
+This document summarizes the current options for integrating Tinfoil/PPQ private models with Routstr, based on the EHBP work in this branch and local testing against `ppq-private-mode-proxy`.
+
+## Background
+
+PPQ private models run behind a Tinfoil/EHBP flow:
+
+- Request bodies are HPKE-encrypted by a Tinfoil client.
+- The PPQ `/private/` endpoint routes ciphertext to the attested enclave.
+- The enclave decrypts, runs inference, and returns an encrypted response.
+- The caller's Tinfoil client decrypts the response locally.
+
+The PPQ private-mode proxy (`~/projects/ppq-private-mode-proxy`) uses this pattern with the JavaScript `tinfoil` SDK:
+
+```ts
+import { SecureClient } from "tinfoil";
+
+const apiBase = "https://api.ppq.ai";
+
+const client = new SecureClient({
+  baseURL: `${apiBase}/private/`,
+  attestationBundleURL: `${apiBase}/private`,
+  transport: "ehbp",
+});
+
+await client.ready();
+
+const response = await client.fetch(`${apiBase}/private/v1/chat/completions`, {
+  method: "POST",
+  headers: {
+    "Content-Type": "application/json",
+    Authorization: `Bearer ${process.env.PPQ_API_KEY}`,
+    "X-Private-Model": "private/gpt-oss-120b",
+    "x-query-source": "api",
+  },
+  body: JSON.stringify({
+    model: "gpt-oss-120b", // enclave-internal model id
+    messages: [{ role: "user", content: "Hello" }],
+  }),
+});
+
+const json = await response.json();
+console.log(json.usage);
+```
+
+`X-Private-Model` carries the PPQ-facing private model id, while the encrypted JSON body uses the enclave-internal model id without the `private/` prefix.
+
+## PPQ private model pricing
+
+PPQ exposes private model pricing through:
+
+```text
+GET https://api.ppq.ai/v1/models?type=all
+```
+
+Filter models whose IDs start with `private/`.
+
+Example private pricing observed:
+
+| Model | Input USD / 1M tokens | Output USD / 1M tokens |
+|---|---:|---:|
+| `private/gpt-oss-120b` | `0.79125` | `1.31875` |
+| `private/llama3-3-70b` | `1.84625` | `2.90125` |
+| `private/qwen3-vl-30b` | `1.31875` | `4.22` |
+| `private/glm-5-2` | `1.5825` | `5.53875` |
+| `private/gemma4-31b` | `0.47475` | `1.055` |
+| `private/kimi-k2-6` | `1.5825` | `5.53875` |
+
+The `ppq-private-mode-proxy` commit `ba984214793d3bca0f7d046b6955d42abc1c6843` changed OpenClaw display metadata to align with a 5% API margin. The live PPQ model endpoint returns the more precise rates above, which already include that margin.
+
+Actual PPQ private billing is:
+
+```text
+price_usd =
+  input_tokens  * input_per_1M_tokens  / 1_000_000
++ output_tokens * output_per_1M_tokens / 1_000_000
+```
+
+A test request to `private/gpt-oss-120b` produced:
+
+```json
+{
+  "input_count": 74,
+  "output_count": 3,
+  "price_in_usd": 0.00006250875
+}
+```
+
+which exactly matches:
+
+```text
+74 * 0.79125 / 1_000_000 + 3 * 1.31875 / 1_000_000
+= 0.00006250875 USD
+```
+
+## Integration architectures
+
+There are three materially different ways Routstr could integrate Tinfoil/PPQ private inference.
+
+## Option A: User integrates Tinfoil directly
+
+```text
+User app / SDK
+  -> Tinfoil SecureClient / TinfoilAI
+    -> EHBP-encrypted request
+      -> Tinfoil/PPQ private enclave
+```
+
+Properties:
+
+- Best privacy for the user.
+- Routstr is not in the request path.
+- User's client encrypts requests and decrypts responses.
+- Usage is visible to the user's app after decryption.
+- Billing is handled directly by Tinfoil/PPQ.
+
+Example with Tinfoil's OpenAI-compatible client:
+
+```ts
+import { TinfoilAI } from "tinfoil";
+
+const client = new TinfoilAI({
+  apiKey: process.env.TINFOIL_API_KEY,
+  transport: "ehbp",
+});
+
+const res = await client.chat.completions.create({
+  model: "llama3-3-70b",
+  messages: [{ role: "user", content: "Hello" }],
+});
+
+console.log(res.choices[0].message.content);
+console.log(res.usage);
+```
+
+This is not a Routstr marketplace flow unless Routstr only acts as discovery/UI around direct Tinfoil/PPQ usage.
+
+## Option B: Routstr integrates Tinfoil as an upstream client
+
+```text
+User -> Routstr plaintext request
+  -> Routstr Tinfoil SecureClient encrypts to PPQ/Tinfoil
+    -> PPQ private enclave
+  -> Routstr receives decrypted response
+  -> Routstr bills from decrypted usage
+  -> Routstr returns plaintext response to user
+```
+
+Properties:
+
+- Easier exact billing.
+- Routstr can read the decrypted OpenAI response and `usage` object.
+- Routstr can charge exact PPQ token pricing.
+- Privacy is different: the user sends plaintext to Routstr, and Routstr sees prompts/responses.
+- End-to-end encryption is only Routstr-to-enclave, not user-to-enclave.
+
+This should be considered a separate product/provider mode, not the same as an end-to-end private relay.
+
+Practical implementation options:
+
+1. Run a Node sidecar that uses the `tinfoil` npm package and expose it as a local HTTP upstream to Routstr.
+2. Port EHBP client behavior to Python.
+3. Reuse or adapt `ppq-private-mode-proxy` as a local upstream.
+
+A Node sidecar is probably the quickest implementation path because `ppq-private-mode-proxy` already demonstrates the full flow.
+
+## Option C: Routstr uses Tinfoil as a direct blind upstream
+
+This is the current branch's design intent and is likely the best fit if Tinfoil/PPQ exposes usage metadata headers:
+
+```text
+User Tinfoil SecureClient
+  -> encrypted request body
+    -> Routstr proxy
+      -> Tinfoil/PPQ private API
+         with X-Tinfoil-Request-Usage-Metrics: true
+        -> PPQ private enclave
+      <- encrypted response
+         plus X-Tinfoil-Usage-Metrics / cost headers
+    <- Routstr proxy
+  -> user decrypts response
+```
+
+In this mode Routstr is still a normal upstream proxy from the user's point of view, but the upstream is Tinfoil/PPQ private inference and the body remains opaque to Routstr.
+
+Properties:
+
+- Strongest privacy with Routstr in the path.
+- Routstr never sees plaintext prompt or plaintext response.
+- Routstr can authenticate and route based on plaintext headers.
+- Routstr should request Tinfoil usage metadata by adding `X-Tinfoil-Request-Usage-Metrics: true` to the upstream request.
+- Tinfoil documents `X-Tinfoil-Usage-Metrics` as an upstream response header for non-streaming requests.
+- For streaming requests, Tinfoil documents `X-Tinfoil-Usage-Metrics` as an HTTP trailer available only after the response body completes.
+- If Tinfoil/PPQ returns `X-Tinfoil-Usage-Metrics` or a cost header on the encrypted response, Routstr can bill exactly without decrypting the body.
+- If usage is only present inside the encrypted response body, Routstr still cannot read it and exact billing is not possible without a separate metadata path.
+
+This is the only architecture that preserves end-to-end encryption from the user to the PPQ/Tinfoil enclave while still letting Routstr mediate payment. The key requirement is that usage/cost metadata must be returned outside the encrypted body, ideally as a response header available before body streaming begins.
+
+## Current Routstr problem
+
+The current EHBP implementation charges successful EHBP requests at `max_cost_for_model` because Routstr cannot decrypt the response body:
+
+```text
+successful EHBP request -> charge full reserved max cost
+```
+
+That is incorrect for PPQ private models. Max cost should be only a reservation/solvency ceiling. Final charge should use actual PPQ private token pricing.
+
+Desired behavior:
+
+```text
+reserve max cost
+forward encrypted request
+obtain actual usage/cost metadata
+finalize actual cost
+refund/release the difference
+```
+
+## Usage/cost metadata requirement
+
+For blind-relay exact billing, PPQ should return one of the following outside the encrypted response body:
+
+```http
+X-PPQ-Cost-USD: 0.00006250875
+```
+
+or:
+
+```http
+X-Private-Usage-Metrics: input=74,output=3
+```
+
+or:
+
+```http
+X-Tinfoil-Usage-Metrics: prompt=74,completion=3,total=77
+```
+
+Tinfoil proxy documentation references a usage-metrics flow where a proxy can request usage via:
+
+```http
+X-Tinfoil-Request-Usage-Metrics: true
+```
+
+and read usage from:
+
+```http
+X-Tinfoil-Usage-Metrics
+```
+
+Docs/example references:
+
+- https://docs.tinfoil.sh/guides/proxy-server
+- https://github.com/tinfoilsh/encrypted-request-proxy-example
+
+During local PPQ testing, PPQ responses included this CORS exposure header:
+
+```http
+Access-Control-Expose-Headers: Ehbp-Response-Nonce, X-Private-Usage-Metrics, X-Encrypted-Usage-Metrics, X-Tinfoil-Usage-Metrics
+```
+
+However, the actual tested non-streaming response did not include any of these usage headers, even when `X-Tinfoil-Request-Usage-Metrics: true` was sent.
+
+The decrypted body did include normal OpenAI usage, but only the decrypting Tinfoil client can see that body.
+
+## Query-history fallback
+
+PPQ's query history endpoint exposes actual usage and cost:
+
+```text
+GET https://api.ppq.ai/queries/history?page=1&page_count=...
+```
+
+A record includes:
+
+```json
+{
+  "timestamp": "...",
+  "model": "private/gpt-oss-120b",
+  "input_count": 74,
+  "output_count": 3,
+  "price_in_usd": 0.00006250875,
+  "query_type": "chat_completion",
+  "query_source": "api"
+}
+```
+
+This could be used as a fallback, but it is less robust than response headers/trailers because matching a request to a history row can be race-prone under concurrency. It would need a reliable request identifier or metadata field that PPQ stores in history.
+
+## Recommended Routstr direction
+
+Use Tinfoil/PPQ as a direct blind upstream and have Routstr explicitly request usage metadata:
+
+```text
+User encrypts body
+Routstr reserves max cost
+Routstr forwards encrypted body to Tinfoil/PPQ /private/
+Routstr includes X-Tinfoil-Request-Usage-Metrics: true
+Tinfoil/PPQ returns X-Tinfoil-Usage-Metrics as a response header for non-streaming,
+or as an HTTP trailer after the body completes for streaming
+Routstr finalizes exact charge
+User decrypts encrypted response
+```
+
+This preserves both:
+
+- privacy: Routstr cannot read prompts/responses;
+- exact billing: Routstr can charge actual PPQ private model cost, assuming Tinfoil/PPQ returns usage/cost metadata outside the encrypted body.
+
+Implementation steps:
+
+1. Update PPQ model fetching to include private models:
+
+   ```text
+   GET https://api.ppq.ai/v1/models?type=all
+   ```
+
+2. Register `private/*` models and any Routstr-facing aliases with correct `forwarded_model_id`.
+
+3. Keep max-cost reservation for bearer keys and X-Cashu solvency checks.
+
+4. Add parsing support for possible usage/cost headers:
+
+   ```http
+   X-PPQ-Cost-USD
+   X-Private-Usage-Metrics
+   X-Tinfoil-Usage-Metrics
+   X-Encrypted-Usage-Metrics
+   ```
+
+5. Finalize by actual cost instead of max cost:
+
+   ```text
+   actual_msats = ceil((actual_usd / sats_usd_price()) * 1000)
+   actual_msats = max(actual_msats, settings.min_request_msat)
+   actual_msats = min(actual_msats, reserved_msats)
+   ```
+
+   or, if only token counts are available:
+
+   ```text
+   actual_usd =
+     input_tokens  * input_per_1M_tokens  / 1_000_000
+   + output_tokens * output_per_1M_tokens / 1_000_000
+   ```
+
+6. If usage/cost metadata is missing, choose an explicit policy:
+
+   - fail closed and refund/revert;
+   - query PPQ history as a fallback;
+   - fallback to max-cost billing only if explicitly configured and clearly disclosed.
+
+Silent max-cost billing should not be the default for PPQ private requests.
+
+## X-Cashu consideration
+
+For bearer-auth requests, finalization can happen after the response stream completes if usage is delivered as a trailer.
+
+For `X-Cashu`, Routstr needs to return the refund token in the response headers. If usage/cost is only available after consuming the encrypted response stream, Routstr may need to buffer EHBP responses before sending them to the client so it can compute the refund amount first.
+
+Possible approaches:
+
+1. Prefer a non-trailer response header with actual cost, available before streaming body starts.
+2. Buffer EHBP X-Cashu responses and then return `X-Cashu` refund.
+3. Introduce a later/refund-claim mechanism, which would be a larger protocol change.
+
+## Summary
+
+- PPQ private models are billed per actual input/output tokens.
+- Private model rates are available from `GET /v1/models?type=all`.
+- Current Routstr EHBP billing at max cost is wrong for PPQ private models.
+- Direct Tinfoil integration inside Routstr would enable exact usage billing but would make Routstr see plaintext.
+- A blind EHBP relay preserves privacy but requires PPQ/Tinfoil to expose usage/cost in plaintext headers/trailers.
+- The preferred solution is to keep Routstr blind and have PPQ return billing metadata outside the encrypted body.
+
+## Implementation status
+
+Direct Tinfoil upstream integration is implemented in `routstr/upstream/tinfoil.py`
+and `routstr/upstream/ehbp.py`.
+
+### What was built
+
+- `TinfoilUpstreamProvider` (`provider_type = "tinfoil"`):
+  - Base URL: `https://inference.tinfoil.sh`
+  - Fetches models from the public `GET /v1/models` endpoint (no auth needed).
+  - Parses Tinfoil's pricing (`inputTokenPricePer1M`, `outputTokenPricePer1M`,
+    `requestPrice`) into the standard `Model`/`Pricing` schema.
+  - `supports_ehbp = True` — acts as a blind EHBP relay.
+  - `get_ehbp_forwarding_target()` returns a target that includes
+    `X-Tinfoil-Request-Usage-Metrics: true`.
+  - `forward_get_request()` proxies `/attestation` to `https://atc.tinfoil.sh/attestation`
+    so the SDK can fetch attestation bundles through Routstr.
+  - Registered in `routstr/upstream/__init__.py` and seeded from
+    `TINFOIL_API_KEY` env var.
+
+- `routstr/upstream/ehbp.py`:
+  - `parse_tinfoil_usage_metrics()` parses
+    `prompt=N,completion=N[,total=N][,model=<name>]` into an OpenAI-style
+    usage dict. The `model` field (added in tinfoilsh/confidential-model-router
+    PR #385) is extracted as a string.
+  - `_resolve_ehbp_target_url()` overrides the forwarding URL with
+    `X-Tinfoil-Enclave-Url` when the SDK sends it.
+  - `_strip_proxy_headers()` removes `X-Routstr-Model`,
+    `X-Tinfoil-Enclave-Url`, and `X-Tinfoil-Request-Usage-Metrics` before
+    forwarding to the enclave.
+  - `_compute_ehbp_actual_cost()` converts the usage header into msats via
+    `calculate_cost()`, clamped to `[min_request_msat, max_cost_for_model]`.
+    When the header's `model=<name>` differs from the requested model, the
+    actual served model's pricing is used for cost calculation.
+  - `forward_ehbp_request()` (bearer auth): if `X-Tinfoil-Usage-Metrics` is
+    present in the response header, finalizes with `adjust_payment_for_tokens()`
+    for exact billing; otherwise falls back to max-cost. Billing uses the
+    actual served model when it differs from the requested one.
+  - `forward_ehbp_x_cashu_request()`: if usage is available, computes the
+    refund from actual cost instead of max cost, using the actual served
+    model's pricing when applicable.
+
+- `routstr/proxy.py`: `/attestation` and `/tee/attestation` paths are forwarded
+  to Tinfoil upstreams without model/cost/auth lookups.
+
+### Billing behavior
+
+| Request shape | Usage source | Billing |
+|---|---|---|
+| Bearer, non-streaming | `X-Tinfoil-Usage-Metrics` response header | Exact token cost via `adjust_payment_for_tokens` |
+| Bearer, streaming | `X-Tinfoil-Usage-Metrics` HTTP trailer | Exact token cost (h11 captures trailers) |
+| Bearer, no usage header/trailer | N/A | Max-cost fallback |
+| X-Cashu, non-streaming | `X-Tinfoil-Usage-Metrics` response header | Refund = `redeemed - actual_cost` |
+| X-Cashu, streaming | `X-Tinfoil-Usage-Metrics` HTTP trailer | Refund = `redeemed - actual_cost` (h11 captures trailers) |
+| X-Cashu, no usage header/trailer | N/A | Refund = `redeemed - max_cost` |
+
+### Cost response headers
+
+Since EHBP response bodies are opaque encrypted blobs, per-request cost cannot
+be injected into the JSON body (as done in the normal proxy flow). Instead,
+Routstr returns cost info as response headers:
+
+| Header | Auth | Description |
+|---|---|---|
+| `X-Routstr-Cost-Msats` | Bearer, X-Cashu | Total msats charged for this request |
+| `X-Routstr-Cost-Usd` | Bearer | USD equivalent of the charge |
+| `X-Routstr-Input-Cost-Msats` | Bearer, X-Cashu | msats attributed to input tokens |
+| `X-Routstr-Output-Cost-Msats` | Bearer, X-Cashu | msats attributed to output tokens |
+
+The client/Tinfoil SDK can read these headers from the HTTP response without
+needing to decrypt the body.
+
+### Setup
+
+```bash
+TINFOIL_API_KEY=your-tinfoil-api-key
+```
+
+The provider is auto-seeded on first startup.
+
+### Usage metrics header format
+
+Tinfoil returns usage metrics in the `X-Tinfoil-Usage-Metrics` response header
+(non-streaming) or HTTP trailer (streaming) when `X-Tinfoil-Request-Usage-Metrics:
+true` is sent. As of tinfoilsh/confidential-model-router PR #385, the format is:
+
+```
+prompt=<prompt_tokens>,completion=<completion_tokens>,total=<total_tokens>,model=<served_model>
+```
+
+The `model` field carries the actual model name served by the enclave.
+Routstr uses this to:
+
+- Verify the served model matches the expected upstream model. The comparison
+  uses ``model_obj.forwarded_model_id`` (the actual upstream ID, e.g.
+  ``glm-5-2``) rather than ``model_obj.id`` (the client-facing alias, e.g.
+  ``tinfoil-glm-5-2``), so aliased models don't trigger a spurious mismatch.
+- When they genuinely differ (Tinfoil served a different upstream model than
+  expected), look up the actual served model's pricing and use it for billing.
+  The reverse lookup uses ``get_model_instance``, which resolves
+  ``forwarded_model_id`` values registered as routable aliases.
+- Log the discrepancy for observability.
+
+If the actual model is not found in Routstr's model registry, billing falls
+back to the requested model's pricing.
+
+### What still needs verification
+
+- ~~End-to-end test with a real Tinfoil SDK client against a Routstr node with
+  `TINFOIL_API_KEY` set.~~ Verified: both non-streaming (header) and streaming
+  (trailer) responses include `model=<name>`.
+- Streaming trailer capture is implemented by buffering the encrypted response
+  in `forward_with_trailer()` and then using the dedicated EHBP payment
+  finalizers for bearer and X-Cashu requests. This provides actual-cost billing
+  today, at the cost of full time-to-last-byte latency for streaming responses.
+- Whether Tinfoil's `/v1/responses` endpoint also returns usage metrics
+  headers or trailers.

logs/.gitkeep

@@ -0,0 +1 @@
+

migrations/versions/02650cd6f028_add_routstr_fees_table.py

@@ -0,0 +1,32 @@
+"""add routstr_fees table
+
+Revision ID: 02650cd6f028
+Revises: c3d4e5f6a7b8
+Create Date: 2026-04-24 00:00:00.000000
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "02650cd6f028"
+down_revision = "c3d4e5f6a7b8"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        "routstr_fees",
+        sa.Column("id", sa.Integer(), nullable=False),
+        sa.Column("accumulated_msats", sa.Integer(), nullable=False, server_default="0"),
+        sa.Column("total_paid_msats", sa.Integer(), nullable=False, server_default="0"),
+        sa.Column("last_paid_at", sa.Integer(), nullable=True),
+        sa.PrimaryKeyConstraint("id"),
+    )
+    # Seed with a single row
+    op.execute("INSERT INTO routstr_fees (id, accumulated_msats, total_paid_msats) VALUES (1, 0, 0)")
+
+
+def downgrade() -> None:
+    op.drop_table("routstr_fees")

migrations/versions/614c0a740e68_add_provider_settings_to_upstream_.py

@@ -0,0 +1,32 @@
+"""add provider_settings to upstream_providers
+
+Revision ID: 614c0a740e68
+Revises: 06f81c0fc88d
+Create Date: 2026-02-13 22:36:53.608737
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "614c0a740e68"
+down_revision = "06f81c0fc88d"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # Check if column exists before adding it
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    columns = [c["name"] for c in inspector.get_columns("upstream_providers")]
+
+    if "provider_settings" not in columns:
+        op.add_column(
+            "upstream_providers",
+            sa.Column("provider_settings", sa.Text(), nullable=True),
+        )
+
+
+def downgrade() -> None:
+    op.drop_column("upstream_providers", "provider_settings")

migrations/versions/a2b3c4d5e6f7_add_constraints_to_lightning_invoices.py

@@ -0,0 +1,26 @@
+"""Add balance_limit, balance_limit_reset, validity_date to lightning_invoices
+
+Revision ID: a2b3c4d5e6f7
+Revises: f1a2b3c4d5e6
+Create Date: 2026-06-03 00:00:00.000000
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "a2b3c4d5e6f7"
+down_revision = "f1a2b3c4d5e6"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column("lightning_invoices", sa.Column("balance_limit", sa.Integer(), nullable=True))
+    op.add_column("lightning_invoices", sa.Column("balance_limit_reset", sa.String(), nullable=True))
+    op.add_column("lightning_invoices", sa.Column("validity_date", sa.Integer(), nullable=True))
+
+
+def downgrade() -> None:
+    op.drop_column("lightning_invoices", "validity_date")
+    op.drop_column("lightning_invoices", "balance_limit_reset")
+    op.drop_column("lightning_invoices", "balance_limit")

migrations/versions/a776ca70e5fe_add_cashu_refunds_table.py

@@ -0,0 +1,42 @@
+"""add cashu_transactions table
+
+Revision ID: a776ca70e5fe
+Revises: 614c0a740e68
+Create Date: 2026-03-11 22:00:01.554762
+"""
+
+import sqlalchemy as sa
+import sqlmodel
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "a776ca70e5fe"
+down_revision = "614c0a740e68"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        "cashu_transactions",
+        sa.Column("id", sqlmodel.sql.sqltypes.AutoString(), nullable=False),
+        sa.Column("token", sqlmodel.sql.sqltypes.AutoString(), nullable=False),
+        sa.Column("amount", sa.Integer(), nullable=False),
+        sa.Column("unit", sqlmodel.sql.sqltypes.AutoString(), nullable=False),
+        sa.Column("mint_url", sqlmodel.sql.sqltypes.AutoString(), nullable=True),
+        sa.Column(
+            "type",
+            sqlmodel.sql.sqltypes.AutoString(),
+            nullable=False,
+            server_default="out",
+        ),
+        sa.Column("request_id", sqlmodel.sql.sqltypes.AutoString(), nullable=True),
+        sa.Column("created_at", sa.Integer(), nullable=False),
+        sa.Column("collected", sa.Boolean(), nullable=False),
+        sa.Column("swept", sa.Boolean(), nullable=False),
+        sa.PrimaryKeyConstraint("id"),
+    )
+
+
+def downgrade() -> None:
+    op.drop_table("cashu_transactions")

migrations/versions/b1c2d3e4f5a6_add_upstream_model_id_to_models.py

@@ -0,0 +1,33 @@
+"""add forwarded_model_id to models
+
+Revision ID: b1c2d3e4f5a6
+Revises: a776ca70e5fe
+Create Date: 2026-04-05 00:00:00.000000
+"""
+
+import sqlalchemy as sa
+import sqlmodel
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "b1c2d3e4f5a6"
+down_revision = "a776ca70e5fe"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "models",
+        sa.Column(
+            "forwarded_model_id",
+            sqlmodel.sql.sqltypes.AutoString(),
+            nullable=True,
+        ),
+    )
+    # Backfill: set forwarded_model_id = id for all existing rows
+    op.execute("UPDATE models SET forwarded_model_id = id WHERE forwarded_model_id IS NULL")
+
+
+def downgrade() -> None:
+    op.drop_column("models", "forwarded_model_id")

migrations/versions/b5e7c9d1f3a2_add_reserved_at_to_api_keys.py

@@ -0,0 +1,25 @@
+"""add reserved_at to api_keys
+
+Revision ID: b5e7c9d1f3a2
+Revises: a2b3c4d5e6f7
+Create Date: 2026-06-12 00:00:00.000000
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "b5e7c9d1f3a2"
+down_revision = "a2b3c4d5e6f7"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # existing keys keep NULL
+    # New reservations populate it via pay_for_request.
+    op.add_column("api_keys", sa.Column("reserved_at", sa.Integer(), nullable=True))
+
+
+def downgrade() -> None:
+    op.drop_column("api_keys", "reserved_at")

migrations/versions/c3d4e5f6a7b8_add_source_to_cashu_transactions.py

@@ -0,0 +1,36 @@
+"""add source to cashu_transactions
+
+Revision ID: c3d4e5f6a7b8
+Revises: b1c2d3e4f5a6
+Create Date: 2026-04-10 00:00:00.000000
+"""
+
+import sqlalchemy as sa
+import sqlmodel
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "c3d4e5f6a7b8"
+down_revision = "b1c2d3e4f5a6"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    columns = [col["name"] for col in inspector.get_columns("cashu_transactions")]
+    if "source" not in columns:
+        op.add_column(
+            "cashu_transactions",
+            sa.Column(
+                "source",
+                sqlmodel.sql.sqltypes.AutoString(),
+                nullable=False,
+                server_default="x-cashu",
+            ),
+        )
+
+
+def downgrade() -> None:
+    op.drop_column("cashu_transactions", "source")

migrations/versions/c6d7e8f9a0b1_add_slug_to_upstream_providers.py

@@ -0,0 +1,88 @@
+"""add slug to upstream_providers
+
+Revision ID: c6d7e8f9a0b1
+Revises: b5e7c9d1f3a2
+Create Date: 2026-06-29 00:00:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+from alembic import op
+
+from routstr.core.provider_slugs import provider_slug_base, provider_slug_candidate
+
+revision = "c6d7e8f9a0b1"
+down_revision = "b5e7c9d1f3a2"
+branch_labels = None
+depends_on = None
+
+
+def _allocate_backfill_slug(provider_type: str, reserved_slugs: set[str]) -> str:
+    base = provider_slug_base(provider_type)
+    suffix_number = 1
+    while True:
+        candidate = provider_slug_candidate(base, suffix_number)
+        if candidate not in reserved_slugs:
+            reserved_slugs.add(candidate)
+            return candidate
+        suffix_number += 1
+
+
+def _backfill_provider_slugs(conn: sa.Connection) -> None:
+    existing_rows = conn.execute(
+        sa.text(
+            "SELECT slug FROM upstream_providers "
+            "WHERE slug IS NOT NULL AND slug != ''"
+        )
+    )
+    reserved_slugs = {str(row.slug).lower() for row in existing_rows}
+
+    rows_to_backfill = conn.execute(
+        sa.text(
+            "SELECT id, provider_type FROM upstream_providers "
+            "WHERE slug IS NULL OR slug = '' "
+            "ORDER BY id"
+        )
+    )
+    for row in rows_to_backfill:
+        slug = _allocate_backfill_slug(str(row.provider_type), reserved_slugs)
+        conn.execute(
+            sa.text("UPDATE upstream_providers SET slug = :slug WHERE id = :id"),
+            {"slug": slug, "id": row.id},
+        )
+
+
+def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    columns = {c["name"] for c in inspector.get_columns("upstream_providers")}
+
+    if "slug" not in columns:
+        op.add_column(
+            "upstream_providers",
+            sa.Column("slug", sa.String(), nullable=True),
+        )
+
+    _backfill_provider_slugs(conn)
+
+    existing_indexes = {idx["name"] for idx in inspector.get_indexes("upstream_providers")}
+    if "ix_upstream_providers_slug" not in existing_indexes:
+        op.create_index(
+            "ix_upstream_providers_slug",
+            "upstream_providers",
+            ["slug"],
+            unique=True,
+        )
+
+
+def downgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    existing_indexes = {idx["name"] for idx in inspector.get_indexes("upstream_providers")}
+    if "ix_upstream_providers_slug" in existing_indexes:
+        op.drop_index("ix_upstream_providers_slug", table_name="upstream_providers")
+
+    columns = {c["name"] for c in inspector.get_columns("upstream_providers")}
+    if "slug" in columns:
+        op.drop_column("upstream_providers", "slug")

migrations/versions/cli_tokens_table.py

@@ -0,0 +1,34 @@
+"""add cli_tokens table
+
+Revision ID: cli_tokens_001
+Revises: e8f9a0b1c2d3
+Create Date: 2026-04-25 00:00:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "cli_tokens_001"
+down_revision = "e8f9a0b1c2d3"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.create_table(
+        "cli_tokens",
+        sa.Column("id", sa.String(), primary_key=True, nullable=False),
+        sa.Column("token", sa.String(), nullable=False, unique=True),
+        sa.Column("name", sa.String(), nullable=False),
+        sa.Column("created_at", sa.Integer(), nullable=False),
+        sa.Column("last_used_at", sa.Integer(), nullable=True),
+        sa.Column("expires_at", sa.Integer(), nullable=True),
+    )
+    op.create_index("ix_cli_tokens_token", "cli_tokens", ["token"], unique=True)
+
+
+def downgrade() -> None:
+    op.drop_index("ix_cli_tokens_token", table_name="cli_tokens")
+    op.drop_table("cli_tokens")

migrations/versions/d4e5f6a7b8c9_add_api_key_to_cashu_transactions.py

@@ -0,0 +1,46 @@
+"""add api key link to cashu_transactions
+
+Revision ID: d4e5f6a7b8c9
+Revises: c3d4e5f6a7b8
+Create Date: 2026-04-20 00:00:00.000000
+"""
+
+import sqlalchemy as sa
+import sqlmodel
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "d4e5f6a7b8c9"
+down_revision = "c3d4e5f6a7b8"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    conn = op.get_bind()
+    inspector = sa.inspect(conn)
+    columns = [col["name"] for col in inspector.get_columns("cashu_transactions")]
+    indexes = {index["name"] for index in inspector.get_indexes("cashu_transactions")}
+
+    if "api_key_hashed_key" not in columns:
+        op.add_column(
+            "cashu_transactions",
+            sa.Column(
+                "api_key_hashed_key",
+                sqlmodel.sql.sqltypes.AutoString(),
+                nullable=True,
+            ),
+        )
+
+    if "ix_cashu_transactions_api_key_hashed_key" not in indexes:
+        op.create_index(
+            "ix_cashu_transactions_api_key_hashed_key",
+            "cashu_transactions",
+            ["api_key_hashed_key"],
+            unique=False,
+        )
+
+
+def downgrade() -> None:
+    op.drop_index("ix_cashu_transactions_api_key_hashed_key", table_name="cashu_transactions")
+    op.drop_column("cashu_transactions", "api_key_hashed_key")

migrations/versions/d7e8f9a0b1c2_add_fee_payout_checkpoint.py

@@ -0,0 +1,37 @@
+"""add fee payout checkpoint
+
+Revision ID: d7e8f9a0b1c2
+Revises: c6d7e8f9a0b1
+Create Date: 2026-07-18 00:00:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+from alembic import op
+
+revision = "d7e8f9a0b1c2"
+down_revision = "c6d7e8f9a0b1"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "routstr_fees",
+        sa.Column(
+            "payout_in_progress_msats",
+            sa.Integer(),
+            nullable=False,
+            server_default="0",
+        ),
+    )
+    op.add_column(
+        "routstr_fees",
+        sa.Column("payout_started_at", sa.Integer(), nullable=True),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("routstr_fees", "payout_started_at")
+    op.drop_column("routstr_fees", "payout_in_progress_msats")

migrations/versions/e8f9a0b1c2d3_merge_heads.py

@@ -0,0 +1,20 @@
+"""merge heads: routstr_fees + api_key_to_cashu_transactions
+
+Revision ID: e8f9a0b1c2d3
+Revises: 02650cd6f028, d4e5f6a7b8c9
+Create Date: 2026-04-24 00:00:00.000000
+"""
+
+# revision identifiers, used by Alembic.
+revision = "e8f9a0b1c2d3"
+down_revision = ("02650cd6f028", "d4e5f6a7b8c9")
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    pass
+
+
+def downgrade() -> None:
+    pass

migrations/versions/f1a2b3c4d5e6_add_created_at_to_api_keys.py

@@ -0,0 +1,25 @@
+"""add created_at to api_keys
+
+Revision ID: f1a2b3c4d5e6
+Revises: cli_tokens_001
+Create Date: 2026-06-01 00:00:00.000000
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "f1a2b3c4d5e6"
+down_revision = "cli_tokens_001"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # Nullable on purpose: existing keys keep NULL (unknown creation time) and
+    # sort last; new keys get populated by the model's default_factory.
+    op.add_column("api_keys", sa.Column("created_at", sa.Integer(), nullable=True))
+
+
+def downgrade() -> None:
+    op.drop_column("api_keys", "created_at")

plans/other-mints-balance-tracking-plan.md

@@ -0,0 +1,355 @@
+# Plan: Track unsupported incoming mints in `other_mints` and include them in balances
+
+## Goal
+
+When a Cashu token arrives from a mint that is **not** in `settings.cashu_mints`, we currently create/use a wallet for that mint and swap value into the primary mint. Any change/surplus left behind after `melt()` remains in the foreign `token_wallet`, but that balance is not surfaced by `fetch_all_balances()` because it only iterates over configured mints.
+
+This plan adds persistent tracking for those foreign mints in a new database table called `other_mints`, and updates balance reporting to include them.
+
+---
+
+## Current behavior
+
+### Incoming unsupported mint flow
+
+In `routstr/wallet.py`:
+
+- `recieve_token()` deserializes the token
+- if `token_obj.mint not in settings.cashu_mints`, it calls `swap_to_primary_mint(token_obj, wallet)`
+- `swap_to_primary_mint()` calls `token_wallet.melt(...)` to pay the primary mint invoice
+
+### Important detail: change is retained, not discarded
+
+The underlying Cashu wallet library keeps any melt change:
+
+- `Wallet.melt()` constructs blank outputs for change
+- when the melt succeeds, returned change is reconstructed into proofs
+- those proofs are appended to `self.proofs` and stored in the wallet DB
+
+So surplus from unsupported mints is **not discarded**, but it may become invisible operationally.
+
+### Visibility problem
+
+`fetch_all_balances()` currently only loops over:
+
+- `settings.cashu_mints`
+- units `sat` and `msat`
+
+This means balances left on unsupported mints are not shown in admin balance reporting.
+
+---
+
+## Proposed design
+
+## 1. Add a new DB table: `other_mints`
+
+Add a small table in `routstr/core/db.py` to persist unsupported mints we have seen in incoming tokens.
+
+Suggested schema:
+
+- `mint_url: str` primary key
+- `created_at: int`
+- `last_seen_at: int`
+
+Minimal model:
+
+```python
+class OtherMint(SQLModel, table=True):
+    __tablename__ = "other_mints"
+
+    mint_url: str = Field(primary_key=True)
+    created_at: int = Field(default_factory=lambda: int(time.time()))
+    last_seen_at: int = Field(default_factory=lambda: int(time.time()))
+```
+
+Why minimal:
+
+- the only required function is mint discovery/tracking
+- unit handling can remain dynamic via existing balance queries over `sat` and `msat`
+
+---
+
+## 2. Add DB helpers for `other_mints`
+
+In `routstr/core/db.py`, add helper functions:
+
+### `register_other_mint(mint_url: str) -> None`
+
+Behavior:
+
+- if the mint is not present, insert it
+- if it already exists, update `last_seen_at`
+
+### `list_other_mints(session) -> list[str]`
+
+Behavior:
+
+- return all tracked unsupported mint URLs
+
+Optional later:
+
+- `delete_other_mint(...)`
+- admin cleanup helpers
+
+---
+
+## 3. Register unsupported mints during token receipt
+
+Update `recieve_token()` in `routstr/wallet.py`.
+
+Current logic:
+
+```python
+if token_obj.mint not in settings.cashu_mints:
+    return await swap_to_primary_mint(token_obj, wallet)
+```
+
+Planned logic:
+
+```python
+if token_obj.mint not in settings.cashu_mints:
+    await db.register_other_mint(token_obj.mint)
+    return await swap_to_primary_mint(token_obj, wallet)
+```
+
+Why here:
+
+- this is the earliest reliable point where we know the mint came in via an actual token
+- this is exactly the path that can leave foreign-mint change behind
+- it avoids needing to infer unsupported mints later from wallet internals
+
+---
+
+## 4. Update `fetch_all_balances()` to include `other_mints`
+
+Current behavior only includes configured mints.
+
+Planned behavior:
+
+- load tracked unsupported mints from DB
+- combine them with `settings.cashu_mints`
+- dedupe while preserving order
+- fetch balances for all tracked mints across requested units
+
+Conceptual flow:
+
+```python
+tracked_mints = dedupe(settings.cashu_mints + other_mints_from_db)
+```
+
+Then existing per-mint/per-unit balance logic can remain mostly unchanged.
+
+This ensures that retained change on unsupported mints becomes visible in admin balance reporting.
+
+---
+
+## 5. Add a balance source marker
+
+Extend `BalanceDetail` in `routstr/wallet.py` to identify whether a balance row comes from a configured mint or an `other_mints` entry.
+
+Suggested field:
+
+- `source: str` with values:
+  - `"configured"`
+  - `"other"`
+
+Updated shape:
+
+```python
+class BalanceDetail(TypedDict, total=False):
+    mint_url: str
+    unit: str
+    source: str
+    wallet_balance: int
+    user_balance: int
+    owner_balance: int
+    error: str
+```
+
+Why this helps:
+
+- admin can distinguish normal configured wallet balances from foreign/unsupported balances
+- avoids confusion if unexpected mint URLs show up in the balances API/UI
+
+---
+
+## 6. Admin/API impact
+
+Backend impact is minimal because `/admin/api/balances` already returns `fetch_all_balances()` output.
+
+Effects:
+
+- supported mints continue to show as before
+- tracked unsupported mints will also appear
+- UI can optionally display the new `source` field
+
+No API contract break is expected if the frontend ignores unknown fields.
+
+---
+
+## 7. Payout behavior: do not change in phase 1
+
+`periodic_payout()` currently only iterates over `settings.cashu_mints`.
+
+Recommendation for this change:
+
+- **do not** expand `periodic_payout()` to include `other_mints` yet
+- only improve visibility through balance reporting
+
+Reason:
+
+- automatic payout from unsupported/foreign mints may be operationally undesirable
+- visibility should come first, automation second
+
+Possible future phase:
+
+- add optional sweeping/payout support for `other_mints`
+- or provide an admin-triggered withdrawal/sweep flow
+
+---
+
+## 8. Logging improvements (optional)
+
+Optional follow-up improvement in `swap_to_primary_mint()`:
+
+- capture the return value from `token_wallet.melt(...)`
+- if feasible, log any reported change amount
+- otherwise, rely on wallet balance reporting to surface residual amounts
+
+This is useful but not required for the first implementation.
+
+---
+
+## Files to change
+
+### `routstr/core/db.py`
+
+Add:
+
+- `OtherMint` SQLModel
+- `register_other_mint()`
+- `list_other_mints()`
+
+### `migrations/versions/<new_revision>_add_other_mints_table.py`
+
+Create migration to add the `other_mints` table.
+
+### `routstr/wallet.py`
+
+Update:
+
+- `recieve_token()` to register unsupported mints
+- `BalanceDetail` to include `source`
+- `fetch_all_balances()` to include both configured and tracked unsupported mints
+
+### `routstr/core/admin.py`
+
+Likely no backend changes required unless a dedicated `other_mints` API is desired.
+
+---
+
+## Behavior rules
+
+### Register a mint when
+
+- an incoming token is processed
+- the token mint is not in `settings.cashu_mints`
+
+### Do not remove automatically when
+
+- balance reaches zero
+
+Reason:
+
+- historical visibility is useful
+- avoids flapping entries in the admin balance list
+- mint may receive additional unsupported tokens later
+
+Potential future enhancement:
+
+- admin endpoint to prune zero-balance `other_mints`
+
+---
+
+## Edge cases
+
+### A mint later becomes configured
+
+If a mint in `other_mints` is later added to `settings.cashu_mints`:
+
+- deduplication prevents duplicate balance rows
+- `source` should resolve to `configured`
+
+### Unsupported mint with zero balance
+
+A tracked unsupported mint may show zero balances.
+
+Initial recommendation:
+
+- allow it to appear
+- consider later filtering zero-balance `other` rows if the UI becomes noisy
+
+### Units
+
+Balance fetching can continue to query both `sat` and `msat` for each tracked mint.
+
+If a mint has no proofs in one unit, current error/zero handling can continue to apply.
+
+---
+
+## Test plan
+
+### DB tests
+
+- registering a new unsupported mint inserts a row
+- registering the same mint again updates `last_seen_at` without duplication
+- listing other mints returns expected mint URLs
+
+### Wallet tests
+
+#### `recieve_token()`
+
+- when mint is unsupported, `db.register_other_mint()` is called before swap
+- when mint is configured, `db.register_other_mint()` is not called
+
+#### `fetch_all_balances()`
+
+- includes configured mints
+- includes `other_mints` from DB
+- dedupes if a mint exists in both configured and other lists
+- sets `source` correctly
+
+### Regression tests
+
+- existing trusted mint balance reporting remains unchanged
+- `/admin/api/balances` continues to work
+
+---
+
+## Recommended implementation order
+
+1. Add `OtherMint` model to `routstr/core/db.py`
+2. Add Alembic migration for `other_mints`
+3. Add `register_other_mint()` and `list_other_mints()` helpers
+4. Update `recieve_token()` to register unsupported mints
+5. Update `fetch_all_balances()` to union configured + tracked other mints
+6. Add `source` to `BalanceDetail`
+7. Add/adjust tests
+
+---
+
+## Summary
+
+This change solves an operational visibility problem:
+
+- unsupported incoming mints can leave retained change in foreign wallets
+- those funds are currently preserved but not surfaced in balance reporting
+- introducing `other_mints` makes those mints discoverable and auditable
+- expanding `fetch_all_balances()` ensures their balances are visible in admin tooling
+
+Recommended scope for the first pass:
+
+- track unsupported mints in DB
+- include them in balance reporting
+- mark them as `source="other"`
+- do not yet change payout/sweeping behavior

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "routstr"
-version = "0.3.0"
+version = "0.4.4"
 description = "Payment proxy for your LLM endpoint using cashu and nostr."
 readme = "README.md"
 requires-python = ">=3.11"
@@ -10,17 +10,18 @@ dependencies = [
     "aiosqlite>=0.20",
     "sqlmodel>=0.0.24",
     "httpx[socks]>=0.25.2",
+    "h11>=0.14",
     "greenlet>=3.2.1",
     "alembic>=1.13",
     "python-json-logger>=2.0.0",
-    "cashu",
-    "secp256k1",
+    "cashu>=0.20",
     "marshmallow>=3.13,<4.0",
     "websockets>=12.0",
     "nostr>=0.0.2",
     "mdurl==0.1.2",
     "pillow>=10",
     "openai>=1.98.0",
+    "litellm>=1.55.0",
 ]
 
 [dependency-groups]
@@ -86,4 +87,3 @@ disallow_untyped_decorators = true
 
 [tool.uv.sources]
 routstr = { workspace = true }
-secp256k1 = { git = "https://github.com/saschanaz/secp256k1-py", branch = "upgrade060" }

routstr/algorithm.py

@@ -86,8 +86,8 @@ def get_provider_penalty(provider: "BaseUpstreamProvider") -> float:
 
 def create_model_mappings(
     upstreams: list["BaseUpstreamProvider"],
-    overrides_by_id: dict[str, tuple],
-    disabled_model_ids: set[str],
+    overrides_by_key: dict[tuple[str, int], tuple],
+    disabled_model_keys: set[tuple[str, int]],
 ) -> tuple[
     dict[str, "Model"], dict[str, list["BaseUpstreamProvider"]], dict[str, "Model"]
 ]:
@@ -107,8 +107,9 @@ def create_model_mappings(
 
     Args:
         upstreams: List of all upstream provider instances
-        overrides_by_id: Dict of model overrides from database {model_id: (ModelRow, fee)}
-        disabled_model_ids: Set of model IDs that should be excluded
+        overrides_by_key: Dict of model overrides from database
+            {(model_id_lower, upstream_provider_id): (ModelRow, fee)}
+        disabled_model_keys: Set of provider-scoped model keys that should be excluded
 
     Returns:
         Tuple of (model_instances, provider_map, unique_models)
@@ -118,12 +119,32 @@ def create_model_mappings(
 
     candidates: dict[str, list[tuple["Model", "BaseUpstreamProvider"]]] = {}
     unique_models: dict[str, "Model"] = {}
+    seen_model_provider: set[tuple[str, str]] = set()
+
+    providers_by_db_id: dict[int, "BaseUpstreamProvider"] = {}
+    for upstream in upstreams:
+        db_id = getattr(upstream, "db_id", None)
+        if isinstance(db_id, int):
+            providers_by_db_id[db_id] = upstream
+
+    # Group upstreams by URL and keep only the one with the lowest fee for each URL
+    upstreams_by_url: dict[str, list["BaseUpstreamProvider"]] = {}
+    for upstream in upstreams:
+        url = getattr(upstream, "base_url", "")
+        if url not in upstreams_by_url:
+            upstreams_by_url[url] = []
+        upstreams_by_url[url].append(upstream)
+
+    filtered_upstreams: list["BaseUpstreamProvider"] = []
+    for providers in upstreams_by_url.values():
+        best_provider = min(providers, key=lambda p: p.provider_fee)
+        filtered_upstreams.append(best_provider)
 
     # Separate OpenRouter from other providers
     openrouter: "BaseUpstreamProvider" | None = None
     other_upstreams: list["BaseUpstreamProvider"] = []
 
-    for upstream in upstreams:
+    for upstream in filtered_upstreams:
         base_url = getattr(upstream, "base_url", "")
         if base_url == "https://openrouter.ai/api/v1":
             openrouter = upstream
@@ -134,6 +155,16 @@ def create_model_mappings(
         """Get base model ID by removing provider prefix."""
         return model_id.split("/", 1)[1] if "/" in model_id else model_id
 
+    def get_provider_identity(upstream: "BaseUpstreamProvider") -> str:
+        """Get a stable provider identity used for deduplication."""
+        db_id = getattr(upstream, "db_id", None)
+        if isinstance(db_id, int):
+            return f"db:{db_id}"
+
+        provider_type = str(getattr(upstream, "provider_type", "") or "").lower()
+        base_url = str(getattr(upstream, "base_url", "") or "").lower()
+        return f"{provider_type}|{base_url}"
+
     def _add_candidate(
         alias: str, model: "Model", provider: "BaseUpstreamProvider"
     ) -> None:
@@ -148,14 +179,23 @@ def create_model_mappings(
     ) -> None:
         """Process all models from a given provider."""
         upstream_prefix = getattr(upstream, "upstream_name", None)
+        provider_key = get_provider_identity(upstream)
+        upstream_db_id = getattr(upstream, "db_id", None)
 
         for model in upstream.get_cached_models():
-            if not model.enabled or model.id in disabled_model_ids:
+            model_key = (
+                (model.id.lower(), upstream_db_id)
+                if isinstance(upstream_db_id, int)
+                else None
+            )
+            if not model.enabled or (
+                model_key is not None and model_key in disabled_model_keys
+            ):
                 continue
 
-            # Apply overrides if present
-            if model.id in overrides_by_id:
-                override_row, provider_fee = overrides_by_id[model.id]
+            # Apply overrides only for this provider's model row.
+            if model_key is not None and model_key in overrides_by_key:
+                override_row, provider_fee = overrides_by_key[model_key]
                 model_to_use = _row_to_model(
                     override_row, apply_provider_fee=True, provider_fee=provider_fee
                 )
@@ -164,14 +204,15 @@ def create_model_mappings(
 
             # Add to unique models
             base_id = get_base_model_id(model_to_use.id)
-            if not is_openrouter or base_id not in unique_models:
+            unique_key = model_to_use.forwarded_model_id or base_id
+            if not is_openrouter or unique_key not in unique_models:
                 unique_model = model_to_use.copy(
                     update={
                         "id": base_id,
                         "upstream_provider_id": upstream.provider_type,
                     }
                 )
-                unique_models[base_id] = unique_model
+                unique_models[unique_key] = unique_model
 
             # Get all aliases for this model
             aliases = resolve_model_alias(
@@ -186,9 +227,14 @@ def create_model_mappings(
                 if prefixed_id not in aliases:
                     aliases.append(prefixed_id)
 
+            # Register forwarded_model_id as a routable alias
+            if model_to_use.forwarded_model_id and model_to_use.forwarded_model_id not in aliases:
+                aliases.append(model_to_use.forwarded_model_id)
+
             # Try to set each alias
             for alias in aliases:
                 _add_candidate(alias, model_to_use, upstream)
+            seen_model_provider.add((model_to_use.id.lower(), provider_key))
 
     # Process non-OpenRouter providers first
     for upstream in other_upstreams:
@@ -198,12 +244,112 @@ def create_model_mappings(
     if openrouter:
         process_provider_models(openrouter, is_openrouter=True)
 
+    # Include enabled DB overrides even when provider discovery misses models.
+    # This is important for deployment-based providers like Azure.
+    for (model_id, upstream_provider_id), override_data in overrides_by_key.items():
+        if (model_id, upstream_provider_id) in disabled_model_keys:
+            continue
+        override_row, provider_fee = override_data
+
+        upstream_for_override = providers_by_db_id.get(upstream_provider_id)
+        if upstream_for_override is None:
+            continue
+
+        provider_key = get_provider_identity(upstream_for_override)
+        dedupe_key = (model_id.lower(), provider_key)
+        if dedupe_key in seen_model_provider:
+            continue
+
+        try:
+            model_to_use = _row_to_model(
+                override_row, apply_provider_fee=True, provider_fee=provider_fee
+            )
+        except Exception as exc:
+            logger.warning(
+                "Skipping invalid model override while building model mappings",
+                extra={
+                    "model_id": model_id,
+                    "upstream_provider_id": upstream_provider_id,
+                    "error": str(exc),
+                    "error_type": type(exc).__name__,
+                },
+            )
+            continue
+        if not model_to_use.enabled:
+            continue
+
+        base_id = get_base_model_id(model_to_use.id)
+        unique_key = model_to_use.forwarded_model_id or base_id
+        is_openrouter = (
+            getattr(upstream_for_override, "base_url", "")
+            == "https://openrouter.ai/api/v1"
+        )
+        if not is_openrouter or unique_key not in unique_models:
+            unique_model = model_to_use.copy(
+                update={
+                    "id": base_id,
+                    "upstream_provider_id": upstream_for_override.provider_type,
+                }
+            )
+            unique_models[unique_key] = unique_model
+
+        try:
+            aliases = resolve_model_alias(
+                model_to_use.id,
+                model_to_use.canonical_slug,
+                alias_ids=model_to_use.alias_ids,
+            )
+        except Exception as exc:
+            logger.warning(
+                "Skipping model aliases for invalid override model",
+                extra={
+                    "model_id": model_id,
+                    "upstream_provider_id": upstream_provider_id,
+                    "error": str(exc),
+                    "error_type": type(exc).__name__,
+                },
+            )
+            continue
+
+        upstream_prefix = getattr(upstream_for_override, "upstream_name", None)
+        if upstream_prefix and "/" not in model_to_use.id:
+            prefixed_id = f"{upstream_prefix}/{model_to_use.id}"
+            if prefixed_id not in aliases:
+                aliases.append(prefixed_id)
+
+        # Register forwarded_model_id as a routable alias
+        if model_to_use.forwarded_model_id and model_to_use.forwarded_model_id not in aliases:
+            aliases.append(model_to_use.forwarded_model_id)
+
+        for alias in aliases:
+            _add_candidate(alias, model_to_use, upstream_for_override)
+        seen_model_provider.add(dedupe_key)
+
     # Sort candidates and build final maps
     model_instances: dict[str, "Model"] = {}
     provider_map: dict[str, list["BaseUpstreamProvider"]] = {}
 
     def alias_priority(model: "Model", alias: str) -> int:
-        """Rank how strong the mapping of alias->model is."""
+        """Rank how strong the mapping of alias->model is.
+
+        forwarded_model_id is the most specific identifier (set per-provider
+        instance), so a match there should beat a model_id match. This way,
+        when multiple providers have the same model_id but different
+        forwarded_model_ids, the one whose forwarded_model_id equals the
+        requested alias wins.
+        """
+        if (
+            model.forwarded_model_id
+            and model.forwarded_model_id.lower() == alias
+        ):
+            return 5
+
+        if (
+            model.id
+            and model.id.lower() == alias
+        ):
+            return 4
+
         model_base = get_base_model_id(model.id)
         if model_base == alias:
             return 3

routstr/auth.py

@@ -7,11 +7,12 @@ from datetime import datetime
 from typing import Optional
 
 from fastapi import HTTPException
+from sqlalchemy import case
 from sqlalchemy.exc import IntegrityError
 from sqlmodel import col, select, update
 
 from .core import get_logger
-from .core.db import ApiKey, AsyncSession
+from .core.db import ApiKey, AsyncSession, accumulate_routstr_fee
 from .core.settings import settings
 from .payment.cost_calculation import (
     CostData,
@@ -19,9 +20,20 @@ from .payment.cost_calculation import (
     MaxCostData,
     calculate_cost,
 )
-from .wallet import credit_balance, deserialize_token_from_string
+from .wallet import (
+    classify_redemption_error,
+    credit_balance,
+    deserialize_token_from_string,
+)
 
 logger = get_logger(__name__)
+payments_logger = get_logger("routstr.payments")
+
+# Routstr platform fee constants
+ROUTSTR_FEE_PERCENT: float = 2.1
+ROUTSTR_LN_ADDRESS: str = "npub130mznv74rxs032peqym6g3wqavh472623mt3z5w73xq9r6qqdufs7ql29s@npub.cash"
+ROUTSTR_FEE_PAYOUT_INTERVAL_SECONDS: int = 900
+ROUTSTR_FEE_DEFAULT_PAYOUT: int = 200
 
 # TODO: implement prepaid api key (not like it was before)
 # PREPAID_API_KEY = os.environ.get("PREPAID_API_KEY", None)
@@ -70,6 +82,37 @@ async def check_and_reset_limit(key: ApiKey, session: AsyncSession) -> bool:
     return False
 
 
+def redemption_error_to_http_exception(error: Exception) -> HTTPException:
+    """Map a Cashu token redemption failure to a sanitized client-facing error.
+
+    Thin wrapper over the shared :func:`classify_redemption_error` so the bearer
+    path stays identical to the X-Cashu and top-up paths.
+    """
+    classified = classify_redemption_error(error)
+    if classified is None:
+        return HTTPException(
+            status_code=500,
+            detail={
+                "error": {
+                    "message": "Internal error during token redemption",
+                    "type": "api_error",
+                    "code": "internal_error",
+                }
+            },
+        )
+    error_type, status_code, message, error_code = classified
+    return HTTPException(
+        status_code=status_code,
+        detail={
+            "error": {
+                "message": message,
+                "type": error_type,
+                "code": error_code,
+            }
+        },
+    )
+
+
 async def validate_bearer_key(
     bearer_key: str,
     session: AsyncSession,
@@ -208,7 +251,17 @@ async def validate_bearer_key(
 
         try:
             hashed_key = hashlib.sha256(bearer_key.encode()).hexdigest()
-            token_obj = deserialize_token_from_string(bearer_key)
+            try:
+                token_obj = deserialize_token_from_string(bearer_key)
+            except Exception as decode_error:
+                # A malformed token is a bad token (400 invalid_cashu_token via
+                # the shared taxonomy), not an auth failure (401) — otherwise it
+                # would fall through to the generic "Invalid API key" handler.
+                raise redemption_error_to_http_exception(
+                    ValueError(
+                        f"Invalid Cashu token: could not decode token ({decode_error})"
+                    )
+                ) from decode_error
             logger.debug(
                 "Generated token hash", extra={"hash_preview": hashed_key[:16] + "..."}
             )
@@ -309,13 +362,13 @@ async def validate_bearer_key(
                 extra={"key_hash": hashed_key[:8] + "..."},
             )
 
-            logger.info(
+            logger.debug(
                 "AUTH: About to call credit_balance",
                 extra={"token_preview": bearer_key[:50]},
             )
             try:
                 msats = await credit_balance(bearer_key, new_key, session)
-                logger.info(
+                logger.debug(
                     "AUTH: credit_balance returned successfully", extra={"msats": msats}
                 )
             except Exception as credit_error:
@@ -326,14 +379,32 @@ async def validate_bearer_key(
                         "error_type": type(credit_error).__name__,
                     },
                 )
-                raise credit_error
+                await session.rollback()
+                raise redemption_error_to_http_exception(credit_error) from credit_error
 
             if msats <= 0:
                 logger.error(
                     "Token redemption returned zero or negative amount",
                     extra={"msats": msats, "key_hash": hashed_key[:8] + "..."},
                 )
-                raise Exception("Token redemption failed")
+                # Defense-in-depth: credit_balance already raises
+                # ValueError("Redeemed token amount must be positive…") before
+                # returning (wallet.py), so this branch is only reachable if a
+                # zero/negative row was somehow persisted; drop it so we never
+                # leave an orphan zero-balance key. Reuse the shared taxonomy
+                # (cashu_error) so the envelope matches the mapper above.
+                await session.delete(new_key)
+                await session.commit()
+                raise HTTPException(
+                    status_code=400,
+                    detail={
+                        "error": {
+                            "message": "Failed to redeem Cashu token: token yielded no value",
+                            "type": "cashu_error",
+                            "code": "cashu_token_zero_value",
+                        }
+                    },
+                )
 
             await session.refresh(new_key)
             await session.commit()
@@ -348,7 +419,10 @@ async def validate_bearer_key(
             )
 
             return new_key
+        except HTTPException:
+            raise
         except Exception as e:
+            await session.rollback()
             logger.error(
                 "Cashu token redemption failed",
                 extra={
@@ -363,19 +437,19 @@ async def validate_bearer_key(
                 status_code=401,
                 detail={
                     "error": {
-                        "message": f"Invalid or expired Cashu key: {str(e)}",
+                        "message": "Invalid or expired Cashu key",
                         "type": "invalid_request_error",
                         "code": "invalid_api_key",
                     }
                 },
             )
 
+    key_preview = bearer_key[:10] + "..." if len(bearer_key) > 10 else bearer_key
     logger.error(
-        "Invalid API key format",
+        f"Invalid API key format: preview={key_preview!r} length={len(bearer_key)} "
+        f"(expected 'sk-...' or 'cashu...' token)",
         extra={
-            "key_preview": bearer_key[:10] + "..."
-            if len(bearer_key) > 10
-            else bearer_key,
+            "key_preview": key_preview,
             "key_length": len(bearer_key),
         },
     )
@@ -384,7 +458,7 @@ async def validate_bearer_key(
         status_code=401,
         detail={
             "error": {
-                "message": "Invalid API key",
+                "message": "Invalid API key format. Expected an 'sk-...' API key or a 'cashu...' token.",
                 "type": "invalid_request_error",
                 "code": "invalid_api_key",
             }
@@ -519,31 +593,19 @@ async def pay_for_request(
     )
 
     # Charge the base cost for the request atomically to avoid race conditions
+    reserved_at_now = int(time.time())
     stmt = (
         update(ApiKey)
         .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
         .where(col(ApiKey.balance) - col(ApiKey.reserved_balance) >= cost_per_request)
         .values(
             reserved_balance=col(ApiKey.reserved_balance) + cost_per_request,
+            reserved_at=reserved_at_now,
             total_requests=col(ApiKey.total_requests) + 1,
         )
     )
     result = await session.exec(stmt)  # type: ignore[call-overload]
 
-    # Also increment total_requests and reserved_balance on the child key if it's different
-    if billing_key.hashed_key != key.hashed_key:
-        child_stmt = (
-            update(ApiKey)
-            .where(col(ApiKey.hashed_key) == key.hashed_key)
-            .values(
-                total_requests=col(ApiKey.total_requests) + 1,
-                reserved_balance=col(ApiKey.reserved_balance) + cost_per_request,
-            )
-        )
-        await session.exec(child_stmt)  # type: ignore[call-overload]
-
-    await session.commit()
-
     if result.rowcount == 0:
         logger.error(
             "Concurrent request depleted balance",
@@ -555,7 +617,6 @@ async def pay_for_request(
             },
         )
 
-        # Another concurrent request spent the balance first
         raise HTTPException(
             status_code=402,
             detail={
@@ -567,6 +628,44 @@ async def pay_for_request(
             },
         )
 
+    # Also increment total_requests and reserved_balance on the child key if it's different.
+    # The balance_limit guard is enforced atomically here — the Python pre-check above
+    # is a fast-path rejection only and provides no concurrency guarantee.
+    if billing_key.hashed_key != key.hashed_key:
+        child_stmt = (
+            update(ApiKey)
+            .where(col(ApiKey.hashed_key) == key.hashed_key)
+            .where(
+                (col(ApiKey.balance_limit).is_(None))
+                | (
+                    col(ApiKey.total_spent)
+                    + col(ApiKey.reserved_balance)
+                    + cost_per_request
+                    <= col(ApiKey.balance_limit)
+                )
+            )
+            .values(
+                total_requests=col(ApiKey.total_requests) + 1,
+                reserved_balance=col(ApiKey.reserved_balance) + cost_per_request,
+                reserved_at=reserved_at_now,
+            )
+        )
+        child_result = await session.exec(child_stmt)  # type: ignore[call-overload]
+
+        if child_result.rowcount == 0:
+            raise HTTPException(
+                status_code=402,
+                detail={
+                    "error": {
+                        "message": f"Balance limit exceeded: {key.balance_limit} mSats limit. {key.total_spent} already spent ({key.reserved_balance} reserved), {cost_per_request} required for this request.",
+                        "type": "insufficient_quota",
+                        "code": "balance_limit_exceeded",
+                    }
+                },
+            )
+
+    await session.commit()
+
     await session.refresh(billing_key)
     if billing_key.hashed_key != key.hashed_key:
         await session.refresh(key)
@@ -582,20 +681,42 @@ async def pay_for_request(
             "total_requests": billing_key.total_requests,
         },
     )
+    payments_logger.info(
+        "RESERVE",
+        extra={
+            "event": "reserve",
+            "key_hash": key.hashed_key[:8] + "...",
+            "billing_key_hash": billing_key.hashed_key[:8] + "...",
+            "cost_reserved": cost_per_request,
+            "balance": billing_key.balance,
+            "reserved_balance": billing_key.reserved_balance,
+            "total_spent": billing_key.total_spent,
+        },
+    )
 
     return cost_per_request
 
 
 async def revert_pay_for_request(
     key: ApiKey, session: AsyncSession, cost_per_request: int
-) -> None:
+) -> bool:
+    """Revert a previously reserved payment. Returns True if revert succeeded,
+    False if the reservation was already released (prevents negative reserved_balance)."""
     billing_key = await get_billing_key(key, session)
 
+    # Keep reserved_at while other reservations remain
+    cleared_reserved_at = case(
+        (col(ApiKey.reserved_balance) - cost_per_request > 0, col(ApiKey.reserved_at)),
+        else_=None,
+    )
+
     stmt = (
         update(ApiKey)
         .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
+        .where(col(ApiKey.reserved_balance) >= cost_per_request)
         .values(
             reserved_balance=col(ApiKey.reserved_balance) - cost_per_request,
+            reserved_at=cleared_reserved_at,
             total_requests=col(ApiKey.total_requests) - 1,
         )
     )
@@ -607,17 +728,19 @@ async def revert_pay_for_request(
         child_stmt = (
             update(ApiKey)
             .where(col(ApiKey.hashed_key) == key.hashed_key)
+            .where(col(ApiKey.reserved_balance) >= cost_per_request)
             .values(
                 total_requests=col(ApiKey.total_requests) - 1,
                 reserved_balance=col(ApiKey.reserved_balance) - cost_per_request,
+                reserved_at=cleared_reserved_at,
             )
         )
         await session.exec(child_stmt)  # type: ignore[call-overload]
 
     await session.commit()
     if result.rowcount == 0:
-        logger.error(
-            "Failed to revert payment - insufficient reserved balance",
+        logger.warning(
+            "Revert skipped - reservation already released (no-op to prevent negative reserved_balance)",
             extra={
                 "key_hash": key.hashed_key[:8] + "...",
                 "billing_key_hash": billing_key.hashed_key[:8] + "...",
@@ -625,28 +748,37 @@ async def revert_pay_for_request(
                 "current_reserved_balance": billing_key.reserved_balance,
             },
         )
-        raise HTTPException(
-            status_code=402,
-            detail={
-                "error": {
-                    "message": f"failed to revert request payment: {cost_per_request} mSats required. {billing_key.balance} available.",
-                    "type": "payment_error",
-                    "code": "payment_error",
-                }
-            },
-        )
+        return False
     await session.refresh(billing_key)
     if billing_key.hashed_key != key.hashed_key:
         await session.refresh(key)
+    payments_logger.info(
+        "REVERT",
+        extra={
+            "event": "revert",
+            "key_hash": key.hashed_key[:8] + "...",
+            "billing_key_hash": billing_key.hashed_key[:8] + "...",
+            "cost_reverted": cost_per_request,
+            "balance": billing_key.balance,
+            "reserved_balance": billing_key.reserved_balance,
+        },
+    )
+    return True
 
 
 async def adjust_payment_for_tokens(
-    key: ApiKey, response_data: dict, session: AsyncSession, deducted_max_cost: int
+    key: ApiKey,
+    response_data: dict,
+    session: AsyncSession,
+    deducted_max_cost: int,
 ) -> dict:
     """
     Adjusts the payment based on token usage in the response.
     This is called after the initial payment and the upstream request is complete.
     Returns cost data to be included in the response.
+
+    The response's usage object is normalized with the default union parser in
+    ``calculate_cost``.
     """
     billing_key = await get_billing_key(key, session)
     model = response_data.get("model", "unknown")
@@ -669,17 +801,19 @@ async def adjust_payment_for_tokens(
             release_stmt = (
                 update(ApiKey)
                 .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
+                .where(col(ApiKey.reserved_balance) >= deducted_max_cost)
                 .values(
                     reserved_balance=col(ApiKey.reserved_balance) - deducted_max_cost
                 )
             )
-            await session.exec(release_stmt)  # type: ignore[call-overload]
+            result = await session.exec(release_stmt)  # type: ignore[call-overload]
 
             # Also release on child key if it's different
             if billing_key.hashed_key != key.hashed_key:
                 child_release_stmt = (
                     update(ApiKey)
                     .where(col(ApiKey.hashed_key) == key.hashed_key)
+                    .where(col(ApiKey.reserved_balance) >= deducted_max_cost)
                     .values(
                         reserved_balance=col(ApiKey.reserved_balance)
                         - deducted_max_cost
@@ -688,14 +822,24 @@ async def adjust_payment_for_tokens(
                 await session.exec(child_release_stmt)  # type: ignore[call-overload]
 
             await session.commit()
-            logger.warning(
-                "Released reservation without charging (fallback)",
-                extra={
-                    "key_hash": key.hashed_key[:8] + "...",
-                    "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                    "deducted_max_cost": deducted_max_cost,
-                },
-            )
+            if result.rowcount == 0:  # type: ignore[union-attr]
+                logger.warning(
+                    "Release reservation skipped - already released (no-op to prevent negative reserved_balance)",
+                    extra={
+                        "key_hash": key.hashed_key[:8] + "...",
+                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                        "deducted_max_cost": deducted_max_cost,
+                    },
+                )
+            else:
+                logger.warning(
+                    "Released reservation without charging (fallback)",
+                    extra={
+                        "key_hash": key.hashed_key[:8] + "...",
+                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                        "deducted_max_cost": deducted_max_cost,
+                    },
+                )
         except Exception as e:
             logger.error(
                 "Failed to release reservation in fallback",
@@ -706,7 +850,18 @@ async def adjust_payment_for_tokens(
                 },
             )
 
-    match await calculate_cost(response_data, deducted_max_cost, session):
+    async def _accumulate_fee(total_cost_msats: int) -> None:
+        if total_cost_msats > 0 and ROUTSTR_FEE_PERCENT > 0:
+            fee_msats = math.ceil(total_cost_msats * ROUTSTR_FEE_PERCENT / 100)
+            try:
+                await accumulate_routstr_fee(session, fee_msats)
+            except Exception as e:
+                logger.warning(
+                    "Failed to accumulate Routstr fee",
+                    extra={"error": str(e), "fee_msats": fee_msats},
+                )
+
+    match await calculate_cost(response_data, deducted_max_cost):
         case MaxCostData() as cost:
             logger.debug(
                 "Using max cost data (no token adjustment)",
@@ -718,11 +873,32 @@ async def adjust_payment_for_tokens(
                 },
             )
             # Finalize by releasing reservation and charging max cost
+            if billing_key.reserved_balance < deducted_max_cost:
+                logger.error(
+                    "reserved_balance below deducted_max_cost before MaxCost finalization — clamping to 0",
+                    extra={
+                        "key_hash": key.hashed_key[:8] + "...",
+                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                        "reserved_balance": billing_key.reserved_balance,
+                        "deducted_max_cost": deducted_max_cost,
+                        "total_cost_msats": cost.total_msats,
+                        "balance": billing_key.balance,
+                        "total_spent": billing_key.total_spent,
+                        "model": model,
+                    },
+                )
+
+            safe_reserved = case(
+                (col(ApiKey.reserved_balance) >= deducted_max_cost,
+                 col(ApiKey.reserved_balance) - deducted_max_cost),
+                else_=0,
+            )
+
             finalize_stmt = (
                 update(ApiKey)
                 .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
                 .values(
-                    reserved_balance=col(ApiKey.reserved_balance) - deducted_max_cost,
+                    reserved_balance=safe_reserved,
                     balance=col(ApiKey.balance) - cost.total_msats,
                     total_spent=col(ApiKey.total_spent) + cost.total_msats,
                 )
@@ -731,13 +907,17 @@ async def adjust_payment_for_tokens(
 
             # Also update total_spent and reserved_balance on the child key if it's different
             if billing_key.hashed_key != key.hashed_key:
+                child_safe_reserved = case(
+                    (col(ApiKey.reserved_balance) >= deducted_max_cost,
+                     col(ApiKey.reserved_balance) - deducted_max_cost),
+                    else_=0,
+                )
                 child_stmt = (
                     update(ApiKey)
                     .where(col(ApiKey.hashed_key) == key.hashed_key)
                     .values(
                         total_spent=col(ApiKey.total_spent) + cost.total_msats,
-                        reserved_balance=col(ApiKey.reserved_balance)
-                        - deducted_max_cost,
+                        reserved_balance=child_safe_reserved,
                     )
                 )
                 await session.exec(child_stmt)  # type: ignore[call-overload]
@@ -766,10 +946,30 @@ async def adjust_payment_for_tokens(
                         "key_hash": key.hashed_key[:8] + "...",
                         "billing_key_hash": billing_key.hashed_key[:8] + "...",
                         "charged_amount": cost.total_msats,
+                        "input_tokens": cost.input_tokens,
+                        "output_tokens": cost.output_tokens,
                         "new_balance": billing_key.balance,
                         "model": model,
                     },
                 )
+                await _accumulate_fee(cost.total_msats)
+                payments_logger.info(
+                    "FINALIZE",
+                    extra={
+                        "event": "finalize",
+                        "key_hash": key.hashed_key[:8] + "...",
+                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                        "model": model,
+                        "cost_reserved": deducted_max_cost,
+                        "cost_charged": cost.total_msats,
+                        "input_tokens": cost.input_tokens,
+                        "output_tokens": cost.output_tokens,
+                        "balance": billing_key.balance,
+                        "reserved_balance": billing_key.reserved_balance,
+                        "total_spent": billing_key.total_spent,
+                        "finalize_type": "max_cost",
+                    },
+                )
             return cost.dict()
 
         case CostData() as cost:
@@ -789,6 +989,8 @@ async def adjust_payment_for_tokens(
                     "cost_difference": cost_difference,
                     "input_msats": cost.input_msats,
                     "output_msats": cost.output_msats,
+                    "input_tokens": cost.input_tokens,
+                    "output_tokens": cost.output_tokens,
                 },
             )
 
@@ -801,12 +1003,32 @@ async def adjust_payment_for_tokens(
                         "model": model,
                     },
                 )
+                if billing_key.reserved_balance < deducted_max_cost:
+                    logger.error(
+                        "reserved_balance below deducted_max_cost on exact-cost finalization — clamping to 0",
+                        extra={
+                            "key_hash": key.hashed_key[:8] + "...",
+                            "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                            "reserved_balance": billing_key.reserved_balance,
+                            "deducted_max_cost": deducted_max_cost,
+                            "total_cost_msats": total_cost_msats,
+                            "balance": billing_key.balance,
+                            "total_spent": billing_key.total_spent,
+                            "model": model,
+                        },
+                    )
+
+                exact_safe_reserved = case(
+                    (col(ApiKey.reserved_balance) >= deducted_max_cost,
+                     col(ApiKey.reserved_balance) - deducted_max_cost),
+                    else_=0,
+                )
+
                 finalize_stmt = (
                     update(ApiKey)
                     .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
                     .values(
-                        reserved_balance=col(ApiKey.reserved_balance)
-                        - deducted_max_cost,
+                        reserved_balance=exact_safe_reserved,
                         balance=col(ApiKey.balance) - total_cost_msats,
                         total_spent=col(ApiKey.total_spent) + total_cost_msats,
                     )
@@ -815,13 +1037,17 @@ async def adjust_payment_for_tokens(
 
                 # Also update total_spent and reserved_balance on the child key if it's different
                 if billing_key.hashed_key != key.hashed_key:
+                    child_exact_safe_reserved = case(
+                        (col(ApiKey.reserved_balance) >= deducted_max_cost,
+                         col(ApiKey.reserved_balance) - deducted_max_cost),
+                        else_=0,
+                    )
                     child_stmt = (
                         update(ApiKey)
                         .where(col(ApiKey.hashed_key) == key.hashed_key)
                         .values(
                             total_spent=col(ApiKey.total_spent) + total_cost_msats,
-                            reserved_balance=col(ApiKey.reserved_balance)
-                            - deducted_max_cost,
+                            reserved_balance=child_exact_safe_reserved,
                         )
                     )
                     await session.exec(child_stmt)  # type: ignore[call-overload]
@@ -830,77 +1056,99 @@ async def adjust_payment_for_tokens(
                 await session.refresh(billing_key)
                 if billing_key.hashed_key != key.hashed_key:
                     await session.refresh(key)
-                return cost.dict()
-
-            # this should never happen why do we handle this???
-            if cost_difference > 0:
-                # Need to charge more than reserved, finalize by releasing reservation and charging total
-                logger.info(
-                    "Additional charge required for token usage",
+                await _accumulate_fee(total_cost_msats)
+                payments_logger.info(
+                    "FINALIZE",
                     extra={
+                        "event": "finalize",
                         "key_hash": key.hashed_key[:8] + "...",
                         "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                        "additional_charge": cost_difference,
-                        "current_balance": billing_key.balance,
-                        "sufficient_balance": billing_key.balance >= cost_difference,
                         "model": model,
+                        "cost_reserved": deducted_max_cost,
+                        "cost_charged": total_cost_msats,
+                        "input_tokens": cost.input_tokens,
+                        "output_tokens": cost.output_tokens,
+                        "balance": billing_key.balance,
+                        "reserved_balance": billing_key.reserved_balance,
+                        "total_spent": billing_key.total_spent,
+                        "finalize_type": "exact",
                     },
                 )
+                return cost.dict()
+
+            # actual cost exceeded discounted reservation (due to tolerance_percentage)
+            if cost_difference > 0:
+                # Always release the reservation and charge min(actual_cost, balance).
+                # CASE expressions keep this atomic and safe even when the
+                # stale-reservation sweeper has already released the reservation.
+                chargeable = case(
+                    (col(ApiKey.balance) >= total_cost_msats, total_cost_msats),
+                    else_=col(ApiKey.balance),
+                )
+                overrun_safe_reserved = case(
+                    (
+                        col(ApiKey.reserved_balance) >= deducted_max_cost,
+                        col(ApiKey.reserved_balance) - deducted_max_cost,
+                    ),
+                    else_=0,
+                )
 
                 finalize_stmt = (
                     update(ApiKey)
                     .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
                     .values(
-                        reserved_balance=col(ApiKey.reserved_balance)
-                        - deducted_max_cost,
-                        balance=col(ApiKey.balance) - total_cost_msats,
-                        total_spent=col(ApiKey.total_spent) + total_cost_msats,
+                        reserved_balance=overrun_safe_reserved,
+                        balance=col(ApiKey.balance) - chargeable,
+                        total_spent=col(ApiKey.total_spent) + chargeable,
                     )
                 )
-                result = await session.exec(finalize_stmt)  # type: ignore[call-overload]
+                await session.exec(finalize_stmt)  # type: ignore[call-overload]
 
-                # Also update total_spent and reserved_balance on the child key if it's different
                 if billing_key.hashed_key != key.hashed_key:
                     child_stmt = (
                         update(ApiKey)
                         .where(col(ApiKey.hashed_key) == key.hashed_key)
                         .values(
-                            total_spent=col(ApiKey.total_spent) + total_cost_msats,
-                            reserved_balance=col(ApiKey.reserved_balance)
-                            - deducted_max_cost,
+                            reserved_balance=overrun_safe_reserved,
+                            total_spent=col(ApiKey.total_spent) + min(billing_key.balance, total_cost_msats),
                         )
                     )
                     await session.exec(child_stmt)  # type: ignore[call-overload]
 
                 await session.commit()
 
-                if result.rowcount:
-                    cost.total_msats = total_cost_msats
-                    await session.refresh(billing_key)
-                    if billing_key.hashed_key != key.hashed_key:
-                        await session.refresh(key)
-
-                    logger.info(
-                        "Finalized payment with additional charge",
-                        extra={
-                            "key_hash": key.hashed_key[:8] + "...",
-                            "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                            "charged_amount": total_cost_msats,
-                            "new_balance": billing_key.balance,
-                            "model": model,
-                        },
-                    )
-                else:
-                    logger.warning(
-                        "Failed to finalize additional charge - releasing reservation",
-                        extra={
-                            "key_hash": key.hashed_key[:8] + "...",
-                            "billing_key_hash": billing_key.hashed_key[:8] + "...",
-                            "attempted_charge": total_cost_msats,
-                            "model": model,
-                        },
-                    )
-                    await release_reservation_only()
+                await session.refresh(billing_key)
+                if billing_key.hashed_key != key.hashed_key:
+                    await session.refresh(key)
+                cost.total_msats = total_cost_msats
+                logger.info(
+                    "Finalized payment with additional charge",
+                    extra={
+                        "key_hash": key.hashed_key[:8] + "...",
+                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                        "charged_amount": total_cost_msats,
+                        "new_balance": billing_key.balance,
+                        "model": model,
+                    },
+                )
+                await _accumulate_fee(total_cost_msats)
+                payments_logger.info(
+                    "FINALIZE",
+                    extra={
+                        "event": "finalize",
+                        "key_hash": key.hashed_key[:8] + "...",
+                        "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                        "model": model,
+                        "cost_reserved": deducted_max_cost,
+                        "cost_charged": total_cost_msats,
+                        "input_tokens": cost.input_tokens,
+                        "output_tokens": cost.output_tokens,
+                        "balance": billing_key.balance,
+                        "reserved_balance": billing_key.reserved_balance,
+                        "total_spent": billing_key.total_spent,
+                        "finalize_type": "overrun",
+                    },
+                )
             else:
                 # Refund some of the base cost
                 refund = abs(cost_difference)
@@ -915,12 +1163,33 @@ async def adjust_payment_for_tokens(
                     },
                 )
 
+                if billing_key.reserved_balance < deducted_max_cost:
+                    logger.error(
+                        "reserved_balance below deducted_max_cost on refund finalization — clamping to 0",
+                        extra={
+                            "key_hash": key.hashed_key[:8] + "...",
+                            "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                            "reserved_balance": billing_key.reserved_balance,
+                            "deducted_max_cost": deducted_max_cost,
+                            "total_cost_msats": total_cost_msats,
+                            "refund_amount": refund,
+                            "balance": billing_key.balance,
+                            "total_spent": billing_key.total_spent,
+                            "model": model,
+                        },
+                    )
+
+                refund_safe_reserved = case(
+                    (col(ApiKey.reserved_balance) >= deducted_max_cost,
+                     col(ApiKey.reserved_balance) - deducted_max_cost),
+                    else_=0,
+                )
+
                 refund_stmt = (
                     update(ApiKey)
                     .where(col(ApiKey.hashed_key) == billing_key.hashed_key)
                     .values(
-                        reserved_balance=col(ApiKey.reserved_balance)
-                        - deducted_max_cost,
+                        reserved_balance=refund_safe_reserved,
                         balance=col(ApiKey.balance) - total_cost_msats,
                         total_spent=col(ApiKey.total_spent) + total_cost_msats,
                     )
@@ -929,13 +1198,17 @@ async def adjust_payment_for_tokens(
 
                 # Also update total_spent and reserved_balance on the child key if it's different
                 if billing_key.hashed_key != key.hashed_key:
+                    child_refund_safe_reserved = case(
+                        (col(ApiKey.reserved_balance) >= deducted_max_cost,
+                         col(ApiKey.reserved_balance) - deducted_max_cost),
+                        else_=0,
+                    )
                     child_stmt = (
                         update(ApiKey)
                         .where(col(ApiKey.hashed_key) == key.hashed_key)
                         .values(
                             total_spent=col(ApiKey.total_spent) + total_cost_msats,
-                            reserved_balance=col(ApiKey.reserved_balance)
-                            - deducted_max_cost,
+                            reserved_balance=child_refund_safe_reserved,
                         )
                     )
                     await session.exec(child_stmt)  # type: ignore[call-overload]
@@ -972,6 +1245,25 @@ async def adjust_payment_for_tokens(
                             "model": model,
                         },
                     )
+                    await _accumulate_fee(total_cost_msats)
+                    payments_logger.info(
+                        "FINALIZE",
+                        extra={
+                            "event": "finalize",
+                            "key_hash": key.hashed_key[:8] + "...",
+                            "billing_key_hash": billing_key.hashed_key[:8] + "...",
+                            "model": model,
+                            "cost_reserved": deducted_max_cost,
+                            "cost_charged": total_cost_msats,
+                            "refunded": refund,
+                            "input_tokens": cost.input_tokens,
+                            "output_tokens": cost.output_tokens,
+                            "balance": billing_key.balance,
+                            "reserved_balance": billing_key.reserved_balance,
+                            "total_spent": billing_key.total_spent,
+                            "finalize_type": "refund",
+                        },
+                    )
 
             return cost.dict()
 
@@ -997,18 +1289,8 @@ async def adjust_payment_for_tokens(
                     }
                 },
             )
-    # Fallback: should not reach here, but release reservation just in case
-    logger.error(
-        "Unexpected fallback in adjust_payment_for_tokens - releasing reservation",
-        extra={"key_hash": key.hashed_key[:8] + "...", "model": model},
-    )
-    await release_reservation_only()
-    return {
-        "base_msats": deducted_max_cost,
-        "input_msats": 0,
-        "output_msats": 0,
-        "total_msats": deducted_max_cost,
-    }
+    # All calculate_cost variants are handled above.
+    raise AssertionError("Unreachable: unhandled calculate_cost result")
 
 
 async def periodic_key_reset() -> None:
@@ -1071,3 +1353,53 @@ async def periodic_key_reset() -> None:
             break
         except Exception as e:
             logger.error(f"Error in periodic_key_reset: {e}")
+
+
+async def periodic_dead_key_prune() -> None:
+    """Periodically prune dead API keys. Interval <= 0 disables it.
+
+    See ``prune_dead_api_keys`` for eligibility.
+    """
+    from .core.db import create_session, prune_dead_api_keys
+
+    interval = settings.dead_key_prune_interval_seconds
+    if interval <= 0:
+        logger.info("Dead-key pruning disabled (interval <= 0)")
+        return
+
+    while True:
+        try:
+            await asyncio.sleep(interval)
+        except asyncio.CancelledError:
+            break
+
+        try:
+            async with create_session() as session:
+                await prune_dead_api_keys(
+                    session, settings.dead_key_min_age_seconds
+                )
+        except asyncio.CancelledError:
+            break
+        except Exception as e:
+            logger.error(f"Error in periodic_dead_key_prune: {e}")
+
+
+STALE_RESERVATION_SWEEP_INTERVAL_SECONDS: int = 60
+
+
+async def periodic_stale_reservation_sweep() -> None:
+    """Background task that releases reservations leaked by client disconnects,
+    crashes or abandoned streams.
+    """
+    from .core.db import create_session, release_stale_reservations
+
+    while True:
+        try:
+            async with create_session() as session:
+                await release_stale_reservations(
+                    session, settings.stale_reservation_timeout_seconds
+                )
+        except Exception:
+            logger.exception("Error in periodic_stale_reservation_sweep")
+
+        await asyncio.sleep(STALE_RESERVATION_SWEEP_INTERVAL_SECONDS)

routstr/balance.py

@@ -5,15 +5,29 @@ from time import monotonic
 from typing import Annotated, NoReturn
 
 from fastapi import APIRouter, Depends, Header, HTTPException
+from fastapi.responses import JSONResponse
 from pydantic import BaseModel
-from sqlmodel import select
+from sqlmodel import col, or_, select, update
 
 from .auth import get_billing_key, validate_bearer_key
-from .core.db import ApiKey, AsyncSession, get_session
+from .core.db import (
+    ApiKey,
+    AsyncSession,
+    CashuTransaction,
+    get_session,
+    store_cashu_transaction,
+)
 from .core.logging import get_logger
 from .core.settings import settings
 from .lightning import lightning_router
-from .wallet import credit_balance, recieve_token, send_to_lnurl, send_token
+from .wallet import (
+    classify_redemption_error,
+    credit_balance,
+    is_mint_connection_error,
+    recieve_token,
+    send_to_lnurl,
+    send_token,
+)
 
 router = APIRouter()
 balance_router = APIRouter(prefix="/v1/balance")
@@ -38,10 +52,9 @@ async def get_balance_info(key: ApiKey, session: AsyncSession) -> dict:
     billing_key = await get_billing_key(key, session)
     info = {
         "api_key": "sk-" + key.hashed_key,
-        "balance": billing_key.balance,
+        "balance": billing_key.total_balance,
         "reserved": billing_key.reserved_balance,
         "is_child": key.parent_key_hash is not None,
-        "parent_key": "sk-" + key.parent_key_hash if key.parent_key_hash else None,
         "total_requests": key.total_requests,
         "total_spent": key.total_spent,
         "balance_limit": key.balance_limit,
@@ -49,7 +62,9 @@ async def get_balance_info(key: ApiKey, session: AsyncSession) -> dict:
         "validity_date": key.validity_date,
     }
 
-    if not key.parent_key_hash:
+    if key.parent_key_hash:
+        info["parent_key_preview"] = key.parent_key_hash[:8] + "..."
+    else:
         # Fetch child keys if this is a parent key
         statement = select(ApiKey).where(ApiKey.parent_key_hash == key.hashed_key)
         results = await session.exec(statement)
@@ -148,16 +163,18 @@ async def topup_wallet_endpoint(
         raise HTTPException(status_code=400, detail="Invalid token format")
     try:
         amount_msats = await credit_balance(cashu_token, billing_key, session)
-    except ValueError as e:
-        error_msg = str(e)
-        if "already spent" in error_msg.lower():
-            raise HTTPException(status_code=400, detail="Token already spent")
-        elif "invalid" in error_msg.lower() or "decode" in error_msg.lower():
-            raise HTTPException(status_code=400, detail="Invalid token format")
-        else:
-            raise HTTPException(status_code=400, detail="Failed to redeem token")
-    except Exception:
-        raise HTTPException(status_code=500, detail="Internal server error")
+    except Exception as e:
+        # Shared taxonomy so top-up matches the bearer/X-Cashu paths (503 for an
+        # unreachable mint, 422 for fee/swap failures, 400 for token faults).
+        classified = classify_redemption_error(e)
+        if classified is None:
+            logger.error(
+                "topup_wallet_endpoint: unhandled error",
+                extra={"error": str(e), "error_type": type(e).__name__},
+            )
+            raise HTTPException(status_code=500, detail="Internal server error")
+        _type, status_code, message, _code = classified
+        raise HTTPException(status_code=status_code, detail=message)
     return {"msats": amount_msats}
 
 
@@ -190,23 +207,112 @@ async def _refund_cache_set(authorization: str, value: dict[str, str]) -> None:
         _refund_cache[key] = (expiry, value)
 
 
-@router.post("/refund")
+async def _lookup_key_no_create(
+    bearer_value: str, session: AsyncSession
+) -> ApiKey | None:
+    """Look up an existing API key without creating one Used by the refund endpoint"""
+    if bearer_value.startswith("sk-"):
+        return await session.get(ApiKey, bearer_value[3:])
+    if bearer_value.startswith("cashu"):
+        hashed = hashlib.sha256(bearer_value.encode()).hexdigest()
+        return await session.get(ApiKey, hashed)
+    return None
+
+
+async def _restore_balance(
+    session: AsyncSession, hashed_key: str, balance: int, reserved_balance: int, mint_url: str
+) -> None:
+    """Restore balance after a failed refund mint attempt."""
+    restore_stmt = (
+        update(ApiKey)
+        .where(col(ApiKey.hashed_key) == hashed_key)
+        .values(
+            balance=col(ApiKey.balance) + balance,
+            reserved_balance=col(ApiKey.reserved_balance) + reserved_balance,
+        )
+    )
+    await session.exec(restore_stmt)  # type: ignore[call-overload]
+    await session.commit()
+    logger.info(
+        "refund_wallet_endpoint: balance restored after mint failure",
+        extra={"hashed_key": hashed_key, "restored_balance": balance, "mint_url": mint_url},
+    )
+
+
+@router.post("/refund", response_model=None)
 async def refund_wallet_endpoint(
-    authorization: Annotated[str, Header(...)],
+    authorization: Annotated[str | None, Header()] = None,
+    x_cashu: Annotated[str | None, Header()] = None,
     session: AsyncSession = Depends(get_session),
-) -> dict[str, str]:
-    if not authorization.startswith("Bearer "):
+) -> JSONResponse | dict[str, str]:
+    if x_cashu:
+        # Find the "in" transaction by the original payment token
+        in_tx_result = await session.exec(
+            select(CashuTransaction).where(
+                CashuTransaction.token == x_cashu,
+                CashuTransaction.type == "in",
+            )
+        )
+        in_tx = in_tx_result.first()
+        if in_tx is None:
+            raise HTTPException(status_code=404, detail="Refund not found")
+
+        # Use the request_id to find the associated "out" (refund) transaction
+        if in_tx.request_id is None:
+            raise HTTPException(status_code=404, detail="Refund not found")
+
+        out_tx_result = await session.exec(
+            select(CashuTransaction).where(
+                CashuTransaction.request_id == in_tx.request_id,
+                CashuTransaction.type == "out",
+            )
+        )
+        out_tx = out_tx_result.first()
+        if out_tx is None:
+            # The "in" row exists with a request_id, but the "out" (refund)
+            # row hasn't been written yet — the upstream request is still in
+            # flight and the refund will be minted once it completes. Tell the
+            # client to retry instead of 404ing permanently (race condition
+            # where /v1/wallet/refund is polled before the refund exists).
+            logger.debug(
+                "refund_wallet_endpoint: refund pending (in row exists, out row not yet created)",
+                extra={"request_id": in_tx.request_id},
+            )
+            raise HTTPException(
+                status_code=425,
+                detail="Refund is pending; retry shortly.",
+                headers={"Retry-After": "2"},
+            )
+        if out_tx.swept:
+            raise HTTPException(status_code=410, detail="Refund has been swept")
+
+        out_tx.collected = True
+        session.add(out_tx)
+        await session.commit()
+        body: dict[str, str] = {"token": out_tx.token}
+        if out_tx.unit == "sat":
+            body["sats"] = str(out_tx.amount)
+        else:
+            body["msats"] = str(out_tx.amount)
+        return JSONResponse(content=body, headers={"X-Cashu": out_tx.token})
+
+    if authorization is None or not authorization.startswith("Bearer "):
         raise HTTPException(
             status_code=401,
             detail="Invalid authorization. Use 'Bearer <cashu-token>' or 'Bearer <api-key>'",
         )
 
     bearer_value: str = authorization[7:]
+    key: ApiKey | None = await _lookup_key_no_create(bearer_value, session)
+    if key is None:
+        raise HTTPException(
+            status_code=401,
+            detail="Key not found. Deposit first via /v1/wallet/create before requesting a refund.",
+        )
 
-    key: ApiKey = await validate_bearer_key(bearer_value, session)
-
-    if cached := await _refund_cache_get(bearer_value):
-        return cached
+    if key.total_balance <= 0:
+        if cached := await _refund_cache_get(bearer_value):
+            return cached
 
     if key.parent_key_hash:
         raise HTTPException(
@@ -214,6 +320,39 @@ async def refund_wallet_endpoint(
             detail="Cannot refund child key. Please refund the parent key instead.",
         )
 
+    if key.reserved_balance > 0:
+        # Release the reservation if it is stale
+        cutoff = int(time.time()) - settings.stale_reservation_timeout_seconds
+        stale_release_stmt = (
+            update(ApiKey)
+            .where(col(ApiKey.hashed_key) == key.hashed_key)
+            .where(col(ApiKey.reserved_balance) > 0)
+            .where(
+                or_(
+                    col(ApiKey.reserved_at).is_(None),
+                    col(ApiKey.reserved_at) < cutoff,
+                )
+            )
+            .values(reserved_balance=0, reserved_at=None)
+        )
+        stale_result = await session.exec(stale_release_stmt)  # type: ignore[call-overload]
+        await session.commit()
+
+        if stale_result.rowcount == 0:
+            raise HTTPException(
+                status_code=400,
+                detail="Cannot refund key. There are ongoing requests for this api key.",
+            )
+
+        await session.refresh(key)
+        logger.warning(
+            "refund_wallet_endpoint: released stale reservation before refund",
+            extra={
+                "hashed_key": key.hashed_key,
+                "stale_timeout_seconds": settings.stale_reservation_timeout_seconds,
+            },
+        )
+
     remaining_balance_msats: int = key.total_balance
 
     if key.refund_currency == "sat":
@@ -226,22 +365,51 @@ async def refund_wallet_endpoint(
     elif remaining_balance <= 0:
         raise HTTPException(status_code=400, detail="No balance to refund")
 
-    # Perform refund operation first, before modifying balance
+    # Capture values before debit — the session may refresh key after commit
+    pre_debit_balance = key.balance
+    pre_debit_reserved = key.reserved_balance
+
+    # --- DEBIT FIRST: atomically zero the balance before minting tokens ---
+    # This prevents the race where a concurrent topup/spend happens between
+    # reading the balance and minting the refund token (double-spend).
+    debit_stmt = (
+        update(ApiKey)
+        .where(col(ApiKey.hashed_key) == key.hashed_key)
+        .where(col(ApiKey.balance) == pre_debit_balance)
+        .where(col(ApiKey.reserved_balance) == pre_debit_reserved)
+        .values(balance=0, reserved_balance=0, reserved_at=None)
+    )
+    debit_result = await session.exec(debit_stmt)  # type: ignore[call-overload]
+    await session.commit()
+
+    if debit_result.rowcount == 0:
+        # Balance changed between read and debit — another request is active
+        raise HTTPException(
+            status_code=409,
+            detail="Balance changed concurrently. Please retry the refund.",
+        )
+
+    # --- MINT: balance is locked at zero, safe to create the refund token ---
+    # Proofs from untrusted mints are swapped to primary_mint on receive.
+    # Use primary_mint unless key.refund_mint_url is an explicitly trusted mint.
+    effective_refund_mint = (
+        key.refund_mint_url
+        if key.refund_mint_url and key.refund_mint_url in settings.cashu_mints
+        else settings.primary_mint
+    )
     try:
         if key.refund_address:
-            from .core.settings import settings as global_settings
-
             await send_to_lnurl(
                 remaining_balance,
                 key.refund_currency or "sat",
-                key.refund_mint_url or global_settings.primary_mint,
+                effective_refund_mint,
                 key.refund_address,
             )
             result = {"recipient": key.refund_address}
         else:
             refund_currency = key.refund_currency or "sat"
             token = await send_token(
-                remaining_balance, refund_currency, key.refund_mint_url
+                remaining_balance, refund_currency, effective_refund_mint
             )
             result = {"token": token}
 
@@ -250,32 +418,105 @@ async def refund_wallet_endpoint(
         else:
             result["msats"] = str(remaining_balance_msats)
 
+        if "token" in result:
+            logger.info(
+                "refund_wallet_endpoint: cashu token issued",
+                extra={
+                    "path": "/v1/wallet/refund",
+                    "token": result["token"],
+                    "amount": remaining_balance,
+                    "currency": key.refund_currency or "sat",
+                },
+            )
+
     except HTTPException:
-        # Re-raise HTTP exceptions (like 400 for balance too small)
+        # Minting failed — restore the debited balance
+        await _restore_balance(session, key.hashed_key, pre_debit_balance, pre_debit_reserved, key.refund_mint_url or "")
         raise
     except Exception as e:
-        # If refund fails, don't modify the database
+        # Minting failed — restore the debited balance
+        await _restore_balance(session, key.hashed_key, pre_debit_balance, pre_debit_reserved, key.refund_mint_url or "")
         error_msg = str(e)
-        if (
-            "mint" in error_msg.lower()
-            or "connection" in error_msg.lower()
-            or isinstance(e, Exception)
-            and "ConnectError" in str(type(e))
-        ):
+        logger.error(
+            "refund_wallet_endpoint: mint/send failed",
+            extra={
+                "error": error_msg,
+                "error_type": type(e).__name__,
+                "hashed_key": key.hashed_key,
+                "remaining_balance": remaining_balance,
+                "refund_currency": key.refund_currency,
+                "refund_mint_url": key.refund_mint_url,
+                "has_refund_address": bool(key.refund_address),
+            },
+        )
+        if is_mint_connection_error(e):
             raise HTTPException(status_code=503, detail="Mint service unavailable")
         else:
             raise HTTPException(status_code=500, detail="Refund failed")
 
     await _refund_cache_set(bearer_value, result)
 
-    key.balance = 0
-    key.reserved_balance = 0
-    session.add(key)
-    await session.commit()
+    if "token" in result:
+        try:
+            await store_cashu_transaction(
+                token=result["token"],
+                amount=remaining_balance,
+                unit=key.refund_currency or "sat",
+                mint_url=key.refund_mint_url,
+                typ="out",
+                collected=False,
+                source="apikey",
+                api_key_hashed_key=key.hashed_key,
+            )
+        except Exception:
+            pass  # store_cashu_transaction already logs
+
+    logger.info(
+        "refund_wallet_endpoint: refund successful",
+        extra={
+            "refunded_msats": remaining_balance_msats,
+            "previous_reserved_balance": key.reserved_balance,
+        },
+    )
 
     return result
 
 
+@router.get("/history")
+async def wallet_history(
+    key: ApiKey = Depends(get_key_from_header),
+    session: AsyncSession = Depends(get_session),
+) -> dict[str, list[dict[str, str | int | bool | None]]]:
+    if key.parent_key_hash:
+        raise HTTPException(
+            status_code=400,
+            detail="Cannot view child key history. Please use the parent key instead.",
+        )
+
+    result = await session.exec(
+        select(CashuTransaction)
+        .where(CashuTransaction.api_key_hashed_key == key.hashed_key)
+        .order_by(col(CashuTransaction.created_at).desc())
+    )
+    transactions = result.all()
+    return {
+        "transactions": [
+            {
+                "id": tx.id,
+                "type": tx.type,
+                "source": tx.source,
+                "amount": tx.amount,
+                "unit": tx.unit,
+                "mint_url": tx.mint_url,
+                "created_at": tx.created_at,
+                "collected": tx.collected,
+                "swept": tx.swept,
+            }
+            for tx in transactions
+        ]
+    }
+
+
 @router.post("/donate")
 async def donate(token: str, ref: str | None = None) -> str:
     try:
@@ -326,10 +567,24 @@ async def create_child_key(
             detail=f"Insufficient balance to create {count} child keys. {total_cost} mSats required.",
         )
 
-    # Deduct cost from parent
-    key.balance -= total_cost
-    key.total_spent += total_cost
-    session.add(key)
+    # Deduct cost from parent atomically — guards against concurrent requests
+    # that both pass the balance check above on stale in-memory state.
+    deduct_stmt = (
+        update(ApiKey)
+        .where(col(ApiKey.hashed_key) == key.hashed_key)
+        .where(col(ApiKey.balance) - col(ApiKey.reserved_balance) >= total_cost)
+        .values(
+            balance=col(ApiKey.balance) - total_cost,
+            total_spent=col(ApiKey.total_spent) + total_cost,
+        )
+    )
+    result = await session.exec(deduct_stmt)  # type: ignore[call-overload]
+
+    if result.rowcount == 0:
+        raise HTTPException(
+            status_code=402,
+            detail=f"Insufficient balance to create {count} child keys. {total_cost} mSats required.",
+        )
 
     # Generate new keys
     import secrets
@@ -354,6 +609,7 @@ async def create_child_key(
         new_keys.append("sk-" + new_key_hash)
 
     await session.commit()
+    await session.refresh(key)
 
     response_data = {
         "api_keys": new_keys,
@@ -400,6 +656,7 @@ async def reset_child_key_spent(
     return {"success": True, "message": "Child key balance reset successfully."}
 
 
+
 @router.api_route(
     "/{path:path}",
     methods=["GET", "POST", "PUT", "DELETE"],
@@ -412,7 +669,7 @@ async def wallet_catch_all(path: str) -> NoReturn:
     )
 
 
-balance_router.include_router(lightning_router)
+balance_router.include_router(lightning_router, include_in_schema=False)
 balance_router.include_router(router)
 
 deprecated_wallet_router = APIRouter(prefix="/v1/wallet", include_in_schema=False)

routstr/core/db.py

@@ -2,14 +2,18 @@ import os
 import pathlib
 import sqlite3
 import time
+import uuid
 from contextlib import asynccontextmanager
 from typing import AsyncGenerator
 
 from alembic import command
 from alembic.config import Config
-from sqlalchemy import UniqueConstraint
+from alembic.util.exc import CommandError
+from sqlalchemy import UniqueConstraint, delete
+from sqlalchemy.exc import OperationalError
 from sqlalchemy.ext.asyncio.engine import create_async_engine
-from sqlmodel import Field, Relationship, SQLModel, func, select, update
+from sqlalchemy.orm import aliased
+from sqlmodel import Field, Relationship, SQLModel, col, func, select, update
 from sqlmodel.ext.asyncio.session import AsyncSession
 
 from .logging import get_logger
@@ -30,6 +34,14 @@ class ApiKey(SQLModel, table=True):  # type: ignore
     reserved_balance: int = Field(
         default=0, description="Reserved balance in millisatoshis (msats)"
     )
+    reserved_at: int | None = Field(
+        default=None,
+        description=(
+            "Unix timestamp of the most recent balance reservation. Used to "
+            "detect and release stale reservations (e.g. after client "
+            "disconnects). NULL when no reservation has been made yet."
+        ),
+    )
     refund_address: str | None = Field(
         default=None,
         description="Lightning address to refund remaining balance after key expires",
@@ -42,6 +54,14 @@ class ApiKey(SQLModel, table=True):  # type: ignore
         default=0, description="Total spent in millisatoshis (msats)"
     )
     total_requests: int = Field(default=0)
+    created_at: int | None = Field(
+        default_factory=lambda: int(time.time()),
+        nullable=True,
+        description=(
+            "Unix timestamp when the key was created. Nullable: keys created "
+            "before this column existed have no value and sort last."
+        ),
+    )
     refund_mint_url: str | None = Field(
         default=None,
         description="URL of the mint used to create the cashu-token",
@@ -76,11 +96,91 @@ class ApiKey(SQLModel, table=True):  # type: ignore
 
 
 async def reset_all_reserved_balances(session: AsyncSession) -> None:
-    logger.info("Resetting all reserved balances to 0")
-    stmt = update(ApiKey).values(reserved_balance=0)
+    stmt = update(ApiKey).values(reserved_balance=0, reserved_at=None)
     await session.exec(stmt)  # type: ignore[call-overload]
     await session.commit()
-    logger.info("Reserved balances reset successfully")
+    logger.info("Reset reserved balances on startup")
+
+
+async def release_stale_reservations(
+    session: AsyncSession, max_age_seconds: int
+) -> int:
+    """Release reservations whose last reserve is older than max_age_seconds.
+    """
+    cutoff = int(time.time()) - max_age_seconds
+    stmt = (
+        update(ApiKey)
+        .where(col(ApiKey.reserved_balance) > 0)
+        .where(col(ApiKey.reserved_at).is_not(None))
+        .where(col(ApiKey.reserved_at) < cutoff)
+        .values(reserved_balance=0, reserved_at=None)
+    )
+    result = await session.exec(stmt)  # type: ignore[call-overload]
+    await session.commit()
+    released = int(result.rowcount or 0)
+    if released:
+        logger.warning(
+            "Released stale balance reservations",
+            extra={"released_keys": released, "max_age_seconds": max_age_seconds},
+        )
+    return released
+
+
+async def prune_dead_api_keys(session: AsyncSession, min_age_seconds: int) -> int:
+    """Delete dead parentless API keys; return the count removed.
+
+    Dead = 0 balance/reservation/spend/requests, older than the grace period,
+    no parent, no children, no pending invoice. Cashu rows are unlinked (not
+    deleted) first to keep the audit trail.
+    """
+    cutoff = int(time.time()) - min_age_seconds
+
+    child = aliased(ApiKey)
+    has_children = (
+        select(child.hashed_key).where(
+            col(child.parent_key_hash) == col(ApiKey.hashed_key)
+        )
+    ).exists()
+    pending_invoice = (
+        select(LightningInvoice.id)
+        .where(col(LightningInvoice.api_key_hash) == col(ApiKey.hashed_key))
+        .where(col(LightningInvoice.status) == "pending")
+    ).exists()
+
+    eligible_hashes = (
+        select(ApiKey.hashed_key)
+        .where(col(ApiKey.balance) == 0)
+        .where(col(ApiKey.reserved_balance) == 0)
+        .where(col(ApiKey.total_spent) == 0)
+        .where(col(ApiKey.total_requests) == 0)
+        .where(col(ApiKey.parent_key_hash).is_(None))
+        .where(
+            (col(ApiKey.created_at).is_(None)) | (col(ApiKey.created_at) < cutoff)
+        )
+        .where(~pending_invoice)
+        .where(~has_children)
+    )
+
+    # Unlink transactions rather than cascade-deleting them, so the financial
+    # audit trail survives. The eligibility predicate is re-evaluated inside both
+    # statements so a key that gained balance mid-run is left untouched.
+    await session.exec(  # type: ignore[call-overload]
+        update(CashuTransaction)
+        .where(col(CashuTransaction.api_key_hashed_key).in_(eligible_hashes))
+        .values(api_key_hashed_key=None)
+    )
+
+    result = await session.exec(  # type: ignore[call-overload]
+        delete(ApiKey).where(col(ApiKey.hashed_key).in_(eligible_hashes))
+    )
+    await session.commit()
+
+    pruned = int(result.rowcount or 0)
+    logger.info(
+        "Pruned dead API keys",
+        extra={"pruned_keys": pruned, "min_age_seconds": min_age_seconds},
+    )
+    return pruned
 
 
 class ModelRow(SQLModel, table=True):  # type: ignore
@@ -103,6 +203,10 @@ class ModelRow(SQLModel, table=True):  # type: ignore
         default=None, description="JSON array of model alias IDs"
     )
     enabled: bool = Field(default=True, description="Whether this model is enabled")
+    forwarded_model_id: str | None = Field(
+        default=None,
+        description="Model ID to use when forwarding requests to upstream provider. Defaults to id if not set.",
+    )
     upstream_provider: "UpstreamProviderRow" = Relationship(back_populates="models")
 
 
@@ -126,6 +230,87 @@ class LightningInvoice(SQLModel, table=True):  # type: ignore
     )
     expires_at: int = Field(description="Unix timestamp when invoice expires")
     paid_at: int | None = Field(default=None, description="Unix timestamp when paid")
+    balance_limit: int | None = Field(
+        default=None,
+        description="Max spendable msats for the created key",
+    )
+    balance_limit_reset: str | None = Field(
+        default=None,
+        description="Reset policy for balance limit (daily, weekly, monthly)",
+    )
+    validity_date: int | None = Field(
+        default=None,
+        description="Unix timestamp after which the created key expires",
+    )
+
+
+class CashuTransaction(SQLModel, table=True):  # type: ignore
+    __tablename__ = "cashu_transactions"
+
+    id: str = Field(
+        primary_key=True,
+        default_factory=lambda: uuid.uuid4().hex,
+        description="Unique transaction identifier",
+    )
+    token: str = Field(description="Serialized Cashu token")
+    amount: int = Field(description="Amount in the token's unit")
+    unit: str = Field(description="Token unit (sat or msat)")
+    mint_url: str | None = Field(default=None, description="Mint URL for the token")
+    type: str = Field(default="out", description="Transaction type: in or out")
+    request_id: str | None = Field(default=None, description="Associated request ID")
+    created_at: int = Field(
+        default_factory=lambda: int(time.time()),
+        description="Unix timestamp",
+    )
+    collected: bool = Field(default=False)
+    swept: bool = Field(default=False)
+    source: str = Field(
+        default="x-cashu",
+        description="Payment source: x-cashu or apikey",
+    )
+    api_key_hashed_key: str | None = Field(
+        default=None,
+        foreign_key="api_keys.hashed_key",
+        index=True,
+        description="Associated API key hash for wallet history",
+    )
+
+
+async def store_cashu_transaction(
+    token: str,
+    amount: int,
+    unit: str,
+    mint_url: str | None = None,
+    typ: str = "out",
+    request_id: str | None = None,
+    collected: bool = False,
+    created_at: int | None = None,
+    source: str = "x-cashu",
+    api_key_hashed_key: str | None = None,
+) -> bool:
+    try:
+        async with create_session() as session:
+            tx = CashuTransaction(
+                token=token,
+                amount=amount,
+                unit=unit,
+                mint_url=mint_url,
+                type=typ,
+                request_id=request_id,
+                collected=collected,
+                created_at=created_at or int(time.time()),
+                source=source,
+                api_key_hashed_key=api_key_hashed_key,
+            )
+            session.add(tx)
+            await session.commit()
+        return True
+    except Exception as e:
+        logger.warning(
+            f"Failed to store cashu transaction: {e} (type={typ})",
+            extra={"error": str(e), "type": typ},
+        )
+        return False
 
 
 class UpstreamProviderRow(SQLModel, table=True):  # type: ignore
@@ -136,6 +321,12 @@ class UpstreamProviderRow(SQLModel, table=True):  # type: ignore
         ),
     )
     id: int | None = Field(default=None, primary_key=True)
+    slug: str | None = Field(
+        default=None,
+        unique=True,
+        index=True,
+        description="Stable external slug used for updates via API key.",
+    )
     provider_type: str = Field(
         description="Provider type: custom, openai, anthropic, azure, openrouter, etc."
     )
@@ -148,12 +339,101 @@ class UpstreamProviderRow(SQLModel, table=True):  # type: ignore
     provider_fee: float = Field(
         default=1.01, description="Provider fee multiplier (default 1%)"
     )
+    provider_settings: str | None = Field(
+        default=None, description="JSON string for provider-specific settings"
+    )
     models: list["ModelRow"] = Relationship(
         back_populates="upstream_provider",
         sa_relationship_kwargs={"cascade": "all, delete-orphan"},
     )
 
 
+class RoutstrFee(SQLModel, table=True):  # type: ignore
+    __tablename__ = "routstr_fees"
+    id: int = Field(default=1, primary_key=True)
+    accumulated_msats: int = Field(default=0)
+    total_paid_msats: int = Field(default=0)
+    last_paid_at: int | None = Field(default=None)
+    payout_in_progress_msats: int = Field(default=0)
+    payout_started_at: int | None = Field(default=None)
+
+
+class CliToken(SQLModel, table=True):  # type: ignore
+    """Long-lived authorization token for CLI/agent use against admin endpoints."""
+
+    __tablename__ = "cli_tokens"
+    id: str = Field(
+        primary_key=True, default_factory=lambda: uuid.uuid4().hex
+    )
+    token: str = Field(unique=True, index=True, description="Bearer token value")
+    name: str = Field(description="Human-readable label for this token")
+    created_at: int = Field(default_factory=lambda: int(time.time()))
+    last_used_at: int | None = Field(default=None)
+    expires_at: int | None = Field(
+        default=None, description="Optional expiry unix timestamp; null = never expires"
+    )
+
+
+async def accumulate_routstr_fee(session: AsyncSession, amount_msats: int) -> None:
+    stmt = (
+        update(RoutstrFee)
+        .where(col(RoutstrFee.id) == 1)
+        .values(accumulated_msats=RoutstrFee.accumulated_msats + amount_msats)
+    )
+    result = await session.exec(stmt)  # type: ignore[call-overload]
+    if result.rowcount == 0:
+        session.add(RoutstrFee(id=1, accumulated_msats=amount_msats))
+    await session.commit()
+
+
+async def get_routstr_fee(session: AsyncSession) -> RoutstrFee:
+    fee = await session.get(RoutstrFee, 1)
+    if fee is None:
+        fee = RoutstrFee(id=1, accumulated_msats=0, total_paid_msats=0)
+        session.add(fee)
+        await session.commit()
+        await session.refresh(fee)
+    return fee
+
+
+async def reset_routstr_fee(session: AsyncSession, paid_msats: int) -> bool:
+    """Checkpoint a fee payout before making the external payment."""
+    stmt = (
+        update(RoutstrFee)
+        .where(col(RoutstrFee.id) == 1)
+        .where(col(RoutstrFee.payout_in_progress_msats) == 0)
+        .where(col(RoutstrFee.accumulated_msats) >= paid_msats)
+        .values(
+            accumulated_msats=RoutstrFee.accumulated_msats - paid_msats,
+            payout_in_progress_msats=paid_msats,
+            payout_started_at=int(time.time()),
+        )
+    )
+    result = await session.exec(stmt)  # type: ignore[call-overload]
+    await session.commit()
+    return result.rowcount == 1
+
+
+async def complete_routstr_fee_payout(
+    session: AsyncSession, paid_msats: int
+) -> bool:
+    """Mark a checkpointed payout complete after the external payment succeeds."""
+    stmt = (
+        update(RoutstrFee)
+        .where(col(RoutstrFee.id) == 1)
+        .where(col(RoutstrFee.payout_in_progress_msats) == paid_msats)
+        .values(
+            payout_in_progress_msats=0,
+            payout_started_at=None,
+            total_paid_msats=RoutstrFee.total_paid_msats + paid_msats,
+            last_paid_at=int(time.time()),
+        )
+    )
+    result = await session.exec(stmt)  # type: ignore[call-overload]
+    await session.commit()
+    return result.rowcount == 1
+
+
 async def balances_for_mint_and_unit(
     db_session: AsyncSession, mint_url: str, unit: str
 ) -> int:
@@ -224,6 +504,17 @@ def fix_cashu_migrations() -> None:
             logger.warning(f"Could not check/fix Cashu database {db_file}: {e}")
 
 
+def _clear_alembic_version() -> None:
+    """Clear the alembic_version table so stamp/upgrade can proceed."""
+    sync_url = DATABASE_URL.replace("+aiosqlite", "")
+    from sqlalchemy import create_engine, text
+
+    eng = create_engine(sync_url)
+    with eng.begin() as conn:
+        conn.execute(text("DELETE FROM alembic_version"))
+    eng.dispose()
+
+
 def run_migrations() -> None:
     """Run Alembic migrations programmatically."""
     try:
@@ -245,8 +536,30 @@ def run_migrations() -> None:
         # Set the database URL in the config
         alembic_cfg.set_main_option("sqlalchemy.url", DATABASE_URL)
 
-        # Run migrations to the latest revision
-        command.upgrade(alembic_cfg, "head")
+        try:
+            command.upgrade(alembic_cfg, "head")
+        except CommandError as e:
+            if "Can't locate revision" in str(e):
+                logger.warning(
+                    "Database stamped with unknown revision (likely from another branch). "
+                    "Re-stamping to current head.",
+                    extra={"error": str(e)},
+                )
+                _clear_alembic_version()
+                command.stamp(alembic_cfg, "head")
+            else:
+                raise
+        except OperationalError as e:
+            if "duplicate column name" in str(e).lower():
+                logger.warning(
+                    "Migration hit a column that already exists (likely added via "
+                    "create_all on another branch). Stamping to current head.",
+                    extra={"error": str(e)},
+                )
+                _clear_alembic_version()
+                command.stamp(alembic_cfg, "head")
+            else:
+                raise
 
         logger.info("Database migrations completed successfully")
 

routstr/core/exceptions.py

@@ -7,11 +7,26 @@ logger = get_logger(__name__)
 
 
 class UpstreamError(Exception):
-    """Exception raised when an upstream provider fails."""
+    """Exception raised when an upstream provider fails.
 
-    def __init__(self, message: str, status_code: int = 502):
+    ``code`` carries a stable, machine-readable classification (e.g.
+    ``UPSTREAM_RATE_LIMIT``) so callers can distinguish failure kinds without
+    string-matching the message. ``details`` holds optional structured,
+    redaction-safe context. Both default to ``None`` for backwards
+    compatibility.
+    """
+
+    def __init__(
+        self,
+        message: str,
+        status_code: int = 502,
+        code: str | None = None,
+        details: dict[str, object] | None = None,
+    ):
         self.message = message
         self.status_code = status_code
+        self.code = code
+        self.details = details
         super().__init__(message)
 
 
@@ -22,16 +37,20 @@ async def http_exception_handler(request: Request, exc: Exception) -> JSONRespon
     # Get status code and detail - works for both FastAPI and Starlette HTTPException
     status_code = getattr(exc, "status_code", 500)
     detail = getattr(exc, "detail", str(exc))
+    path = request.url.path
 
-    logger.warning(
-        "HTTP exception",
-        extra={
-            "request_id": request_id,
-            "status_code": status_code,
-            "detail": detail,
-            "path": request.url.path,
-        },
-    )
+    # 4xx is client behaviour; the uvicorn access log already records it.
+    # Only 5xx warrants a server-side warning/error log here.
+    if status_code >= 500:
+        logger.error(
+            f"HTTP {status_code} on {path}: {detail}",
+            extra={
+                "request_id": request_id,
+                "status_code": status_code,
+                "detail": detail,
+                "path": path,
+            },
+        )
 
     return JSONResponse(
         status_code=status_code,

routstr/core/logging.py

@@ -3,50 +3,63 @@ Logging configuration for Routstr.
 
 CRITICAL LOG MESSAGES FOR USAGE STATISTICS:
 ===========================================
-The following log messages are parsed by the usage tracking system (routstr/core/admin.py).
+The following log messages are parsed by the usage tracking system
+(routstr/core/usage_analytics_store.py and routstr/core/log_manager.py).
 DO NOT modify or remove these messages without updating the usage tracking logic:
 
 1. "Received proxy request" (INFO) - routstr/proxy.py
    - Used to count total incoming requests
    - Includes model information in context
 
-    2. "Payment adjustment completed for streaming" (INFO) - routstr/upstream/base.py
-       "Payment adjustment completed for non-streaming" (INFO) - routstr/upstream/base.py
+2. "Calculated token-based cost" (INFO) - routstr/auth.py
    - Used to track successful completions and revenue
-   - The 'cost_data.total_msats' field is extracted for revenue calculation
-   - Must include 'cost_data' in extra dict
+   - The 'token_cost', 'model', 'input_tokens', and 'output_tokens' fields are extracted for dashboard metrics
 
-3. "Payment processed successfully" (INFO) - routstr/auth.py
+3. "Max cost payment finalized" (INFO) - routstr/auth.py
+   - Used as the successful completion fallback when token usage is unavailable
+   - The 'charged_amount', 'model', 'input_tokens', and 'output_tokens' fields are extracted for dashboard metrics
+
+4. "Payment processed successfully" (INFO) - routstr/auth.py
    - Used to count successful payment processing events
    - Tracks payment-related metrics
 
-4. "Upstream request failed, revert payment" (WARNING) - routstr/proxy.py
+5. "Upstream request failed, revert payment" (WARNING) - routstr/proxy.py
    - Used to track failed requests and refunds
    - The 'max_cost_for_model' field is extracted for refund calculation
    - Must include 'max_cost_for_model' in extra dict
 
-5. Any ERROR level logs with "upstream" in the message
+6. Any ERROR level logs with "upstream" in the message
    - Used to count upstream provider errors
    - Helps identify service reliability issues
 
 If you need to modify these messages, ensure you also update the parsing logic in:
-- routstr/core/admin.py:_aggregate_metrics_by_time()
-- routstr/core/admin.py:_get_summary_stats()
-- routstr/core/admin.py:get_revenue_by_model()
+- routstr/core/usage_analytics_store.py
+- routstr/core/log_manager.py
 """
 
 import logging.config
 import logging.handlers
 import os
 import re
+import sys
 import tomllib
 from datetime import datetime
 from pathlib import Path
 from typing import Any
 
 from pythonjsonlogger import jsonlogger
+from rich.console import Console
 from rich.logging import RichHandler
 
+from .redaction import redact_obj, redact_org_ids
+
+# Only use RichHandler when stdout is a real TTY. In non-TTY contexts
+# (docker logs, pipes, CI) Rich pads every line to width and wraps long
+# records, producing visually-empty trailing whitespace and split records.
+# A plain StreamHandler avoids both problems.
+_stdout_is_tty = sys.stdout.isatty()
+_console = Console(soft_wrap=True) if _stdout_is_tty else None
+
 # Define custom TRACE level
 TRACE_LEVEL = 5
 logging.addLevelName(TRACE_LEVEL, "TRACE")
@@ -169,6 +182,37 @@ class RequestIdFilter(logging.Filter):
         return True
 
 
+# Standard ``LogRecord`` attributes that are never user-supplied ``extra``
+# fields; skipped when redacting structured extras (``msg``/``message`` are
+# handled separately above).
+_NON_EXTRA_RECORD_ATTRS = frozenset(
+    {
+        "name",
+        "msg",
+        "args",
+        "levelname",
+        "levelno",
+        "pathname",
+        "filename",
+        "module",
+        "exc_info",
+        "exc_text",
+        "stack_info",
+        "lineno",
+        "funcName",
+        "created",
+        "msecs",
+        "relativeCreated",
+        "thread",
+        "threadName",
+        "processName",
+        "process",
+        "taskName",
+        "message",
+    }
+)
+
+
 class SecurityFilter(logging.Filter):
     """Filter to remove sensitive information from logs."""
 
@@ -192,6 +236,7 @@ class SecurityFilter(logging.Filter):
         """Filter out sensitive information from log records."""
         try:
             message = record.getMessage()
+            message = redact_org_ids(message)
             standalone_patterns = [
                 r"Bearer\s+([a-zA-Z0-9_\-\.]{10,})",  # Bearer token (must be 10 characters or more to reduce false-positives)
                 r"cashu[A-Z]+([a-zA-Z0-9_\-\.=/+]+)",  # Cashu tokens
@@ -213,6 +258,16 @@ class SecurityFilter(logging.Filter):
             record.msg = message
             record.args = ()
 
+            # Structured `extra={...}` fields are emitted by the JSON formatter
+            # straight from the record dict and never pass through the message
+            # formatting above. Redact organization IDs from any string-valued
+            # extra so they cannot leak via structured logs.
+            for attr, value in list(record.__dict__.items()):
+                if attr in _NON_EXTRA_RECORD_ATTRS:
+                    continue
+                if isinstance(value, (str, dict, list, tuple)):
+                    record.__dict__[attr] = redact_obj(value)
+
         except Exception:
             pass
 
@@ -259,6 +314,26 @@ def setup_logging() -> None:
     if console_enabled:
         handlers.append("console")
 
+    if _stdout_is_tty:
+        console_handler: dict[str, Any] = {
+            "()": RichHandler,
+            "level": log_level,
+            "show_time": False,
+            "show_path": False,
+            "rich_tracebacks": True,
+            "markup": True,
+            "console": _console,
+            "filters": ["request_id_filter", "security_filter"],
+        }
+    else:
+        console_handler = {
+            "class": "logging.StreamHandler",
+            "level": log_level,
+            "formatter": "plain",
+            "stream": "ext://sys.stdout",
+            "filters": ["request_id_filter", "security_filter"],
+        }
+
     LOGGING_CONFIG = {
         "version": 1,
         "disable_existing_loggers": False,
@@ -268,6 +343,10 @@ def setup_logging() -> None:
                 "format": "%(asctime)s %(name)s %(levelname)s %(message)s %(pathname)s %(lineno)d %(version)s %(request_id)s",
                 "datefmt": "%Y-%m-%d %H:%M:%S",
             },
+            "plain": {
+                "format": "%(asctime)s %(levelname)-7s %(name)s %(message)s",
+                "datefmt": "%Y-%m-%d %H:%M:%S",
+            },
         },
         "filters": {
             "version_filter": {"()": VersionFilter},
@@ -275,15 +354,7 @@ def setup_logging() -> None:
             "security_filter": {"()": SecurityFilter},
         },
         "handlers": {
-            "console": {
-                "()": RichHandler,
-                "level": log_level,
-                "show_time": False,
-                "show_path": False,
-                "rich_tracebacks": True,
-                "markup": True,
-                "filters": ["request_id_filter", "security_filter"],
-            },
+            "console": console_handler,
             "file": {
                 "()": DailyRotatingFileHandler,
                 "level": log_level,

routstr/core/main.py

@@ -1,5 +1,4 @@
 import asyncio
-import os
 from contextlib import asynccontextmanager
 from pathlib import Path
 from typing import AsyncGenerator
@@ -9,32 +8,43 @@ from fastapi.middleware.cors import CORSMiddleware
 from fastapi.responses import FileResponse, RedirectResponse
 from fastapi.staticfiles import StaticFiles
 from starlette.exceptions import HTTPException
+from starlette.responses import Response as StarletteResponse
+from starlette.types import Scope
 
-from ..auth import periodic_key_reset
+from ..auth import (
+    periodic_dead_key_prune,
+    periodic_key_reset,
+    periodic_stale_reservation_sweep,
+)
 from ..balance import balance_router, deprecated_wallet_router
-from ..nostr import announce_provider, providers_cache_refresher
+from ..lightning import lightning_router, periodic_invoice_watcher
+from ..nostr import (
+    announce_provider,
+    providers_cache_refresher,
+    publish_usage_analytics,
+)
 from ..nostr.discovery import providers_router
 from ..payment.models import models_router, update_sats_pricing
 from ..payment.price import update_prices_periodically
 from ..proxy import initialize_upstreams, proxy_router, refresh_model_maps_periodically
-from ..wallet import periodic_payout
+from ..upstream.auto_topup import periodic_auto_topup
+from ..upstream.deepseek_v4_pricing_shim import register_deepseek_v4_pricing
+from ..upstream.litellm_routing import configure_litellm
+from ..wallet import periodic_payout, periodic_refund_sweep, periodic_routstr_fee_payout
 from .admin import admin_router
 from .db import create_session, init_db, run_migrations
 from .exceptions import general_exception_handler, http_exception_handler
 from .logging import get_logger, setup_logging
 from .middleware import LoggingMiddleware
+from .not_found import _NOT_FOUND_HTML, not_found_catch_all  # noqa: F401
 from .settings import SettingsService
 from .settings import settings as global_settings
+from .version import __version__
 
 # Initialize logging first
 setup_logging()
 logger = get_logger(__name__)
 
-if os.getenv("VERSION_SUFFIX") is not None:
-    __version__ = f"0.3.0-{os.getenv('VERSION_SUFFIX')}"
-else:
-    __version__ = "0.3.0"
-
 
 @asynccontextmanager
 async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
@@ -44,12 +54,28 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
     pricing_task = None
     payout_task = None
     nip91_task = None
+    analytics_task = None
     providers_task = None
     models_refresh_task = None
     model_maps_refresh_task = None
     key_reset_task = None
+    stale_reservation_task = None
+    dead_key_prune_task = None
+    auto_topup_task = None
+    refund_sweep_task = None
+    routstr_fee_task = None
+    invoice_watcher_task = None
 
     try:
+        # Apply litellm-wide settings (drop_params, chat-completions URL,
+        # debug logging) before any upstream provider dispatches a request.
+        configure_litellm()
+
+        # TEMPORARY: backfill DeepSeek V4 pricing missing from litellm's cost
+        # map (BerriAI/litellm#30430). Remove this call and
+        # deepseek_v4_pricing_shim.py once litellm ships these models.
+        register_deepseek_v4_pricing()
+
         # Run database migrations on startup
         run_migrations()
 
@@ -94,16 +120,28 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
         btc_price_task = asyncio.create_task(update_prices_periodically())
         pricing_task = asyncio.create_task(update_sats_pricing())
         if global_settings.models_refresh_interval_seconds > 0:
+            # Pass the accessor (not its current value) so the loop sees providers
+            # added/changed via reinitialize_upstreams() instead of staying pinned
+            # to the startup snapshot.
             models_refresh_task = asyncio.create_task(
-                refresh_upstreams_models_periodically(get_upstreams())
+                refresh_upstreams_models_periodically(get_upstreams)
             )
         model_maps_refresh_task = asyncio.create_task(refresh_model_maps_periodically())
         payout_task = asyncio.create_task(periodic_payout())
         if global_settings.nsec:
             nip91_task = asyncio.create_task(announce_provider())
+        analytics_task = asyncio.create_task(publish_usage_analytics())
         if global_settings.providers_refresh_interval_seconds > 0:
             providers_task = asyncio.create_task(providers_cache_refresher())
         key_reset_task = asyncio.create_task(periodic_key_reset())
+        stale_reservation_task = asyncio.create_task(
+            periodic_stale_reservation_sweep()
+        )
+        dead_key_prune_task = asyncio.create_task(periodic_dead_key_prune())
+        auto_topup_task = asyncio.create_task(periodic_auto_topup())
+        refund_sweep_task = asyncio.create_task(periodic_refund_sweep())
+        routstr_fee_task = asyncio.create_task(periodic_routstr_fee_payout())
+        invoice_watcher_task = asyncio.create_task(periodic_invoice_watcher())
 
         yield
 
@@ -127,6 +165,8 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
             payout_task.cancel()
         if nip91_task is not None:
             nip91_task.cancel()
+        if analytics_task is not None:
+            analytics_task.cancel()
         if providers_task is not None:
             providers_task.cancel()
         if models_refresh_task is not None:
@@ -135,6 +175,18 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
             model_maps_refresh_task.cancel()
         if key_reset_task is not None:
             key_reset_task.cancel()
+        if stale_reservation_task is not None:
+            stale_reservation_task.cancel()
+        if dead_key_prune_task is not None:
+            dead_key_prune_task.cancel()
+        if auto_topup_task is not None:
+            auto_topup_task.cancel()
+        if refund_sweep_task is not None:
+            refund_sweep_task.cancel()
+        if routstr_fee_task is not None:
+            routstr_fee_task.cancel()
+        if invoice_watcher_task is not None:
+            invoice_watcher_task.cancel()
 
         try:
             tasks_to_wait = []
@@ -146,6 +198,8 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
                 tasks_to_wait.append(payout_task)
             if nip91_task is not None:
                 tasks_to_wait.append(nip91_task)
+            if analytics_task is not None:
+                tasks_to_wait.append(analytics_task)
             if providers_task is not None:
                 tasks_to_wait.append(providers_task)
             if models_refresh_task is not None:
@@ -154,6 +208,18 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
                 tasks_to_wait.append(model_maps_refresh_task)
             if key_reset_task is not None:
                 tasks_to_wait.append(key_reset_task)
+            if stale_reservation_task is not None:
+                tasks_to_wait.append(stale_reservation_task)
+            if dead_key_prune_task is not None:
+                tasks_to_wait.append(dead_key_prune_task)
+            if auto_topup_task is not None:
+                tasks_to_wait.append(auto_topup_task)
+            if refund_sweep_task is not None:
+                tasks_to_wait.append(refund_sweep_task)
+            if routstr_fee_task is not None:
+                tasks_to_wait.append(routstr_fee_task)
+            if invoice_watcher_task is not None:
+                tasks_to_wait.append(invoice_watcher_task)
 
             if tasks_to_wait:
                 await asyncio.gather(*tasks_to_wait, return_exceptions=True)
@@ -165,6 +231,23 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
             )
 
 
+class _ImmutableStaticFiles(StaticFiles):
+    """Static files with long Cache-Control for content-hashed Next.js assets.
+
+    Files under `/_next/static/` are emitted with content hashes in their
+    filenames and never mutate, so we serve them with a one-year immutable
+    cache header so browsers and CDNs stop revalidating on every reload.
+    """
+
+    async def get_response(self, path: str, scope: Scope) -> StarletteResponse:
+        response = await super().get_response(path, scope)
+        if response.status_code == 200:
+            response.headers["Cache-Control"] = (
+                "public, max-age=31536000, immutable"
+            )
+        return response
+
+
 app = FastAPI(version=__version__, lifespan=lifespan)
 
 
@@ -174,7 +257,20 @@ app.add_middleware(
     allow_credentials=True,
     allow_methods=["*"],
     allow_headers=["*"],
-    expose_headers=["x-routstr-request-id"],
+    expose_headers=[
+        "x-routstr-request-id",
+        "x-cashu",
+        "x-routstr-cost-msats",
+        "x-routstr-cost-usd",
+        "x-routstr-input-cost-msats",
+        "x-routstr-output-cost-msats",
+        # EHBP (Tinfoil) protocol headers must be exposed so browser clients
+        # can detect and decrypt encrypted responses. Without these, the
+        # browser hides them via CORS and the SDK treats the response as a
+        # plaintext proxy error, returning raw ciphertext.
+        "Ehbp-Response-Nonce",
+        "Ehbp-Encapsulated-Key",
+    ],
 )
 
 # Add logging middleware
@@ -211,7 +307,7 @@ if UI_DIST_PATH.exists() and UI_DIST_PATH.is_dir():
 
     app.mount(
         "/_next",
-        StaticFiles(directory=UI_DIST_PATH / "_next", check_dir=True),
+        _ImmutableStaticFiles(directory=UI_DIST_PATH / "_next", check_dir=True),
         name="next-static",
     )
 
@@ -219,100 +315,70 @@ if UI_DIST_PATH.exists() and UI_DIST_PATH.is_dir():
     async def serve_root_ui() -> FileResponse:
         return FileResponse(UI_DIST_PATH / "index.html")
 
-    # Add explicit route for /index.txt to redirect to /
+    # Serve the App Router RSC payload for the home page.
     @app.get("/index.txt", include_in_schema=False)
-    async def redirect_index_txt() -> RedirectResponse:
-        return RedirectResponse("/")
+    async def serve_root_rsc() -> FileResponse:
+        return FileResponse(
+            UI_DIST_PATH / "index.txt", media_type="text/x-component"
+        )
+
+    # Next.js is built with `trailingSlash: true`, so all UI page URLs end
+    # with a slash (e.g. `/login/`). The proxy router catches `/{path:path}`
+    # before FastAPI's `redirect_slashes` logic can normalize the URL, so we
+    # must register both the with-slash and without-slash variants here.
+    UI_PAGES = (
+        "dashboard",
+        "login",
+        "model",
+        "providers",
+        "settings",
+        "transactions",
+        "balances",
+        "logs",
+        "usage",
+        "unauthorized",
+    )
+
+    def _register_ui_page(name: str) -> None:
+        page_dir = UI_DIST_PATH / name
+        index_html = page_dir / "index.html"
+        index_txt = page_dir / "index.txt"
+
+        async def serve_page() -> FileResponse:
+            return FileResponse(index_html)
+
+        async def serve_page_rsc() -> FileResponse:
+            return FileResponse(index_txt, media_type="text/x-component")
+
+        app.add_api_route(
+            f"/{name}",
+            serve_page,
+            methods=["GET"],
+            include_in_schema=False,
+            name=f"serve_{name}_ui",
+        )
+        app.add_api_route(
+            f"/{name}/",
+            serve_page,
+            methods=["GET"],
+            include_in_schema=False,
+            name=f"serve_{name}_ui_slash",
+        )
+        app.add_api_route(
+            f"/{name}/index.txt",
+            serve_page_rsc,
+            methods=["GET"],
+            include_in_schema=False,
+            name=f"serve_{name}_rsc",
+        )
+
+    for _page in UI_PAGES:
+        _register_ui_page(_page)
 
     @app.get("/admin")
     async def admin_redirect() -> FileResponse:
         return FileResponse(UI_DIST_PATH / "index.html")
 
-    @app.get("/dashboard", include_in_schema=False)
-    async def serve_dashboard_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "index.html")
-
-    @app.get("/login", include_in_schema=False)
-    async def serve_login_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "login" / "index.html")
-
-    # Add explicit route for /login/index.txt to redirect to /login
-    @app.get("/login/index.txt", include_in_schema=False)
-    async def redirect_login_index_txt() -> RedirectResponse:
-        return RedirectResponse("/login")
-
-    @app.get("/model", include_in_schema=False)
-    async def serve_models_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "model" / "index.html")
-
-    # Add explicit route for /model/index.txt to redirect to /model
-    @app.get("/model/index.txt", include_in_schema=False)
-    async def redirect_model_index_txt() -> RedirectResponse:
-        return RedirectResponse("/model")
-
-    @app.get("/providers", include_in_schema=False)
-    async def serve_providers_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "providers" / "index.html")
-
-    # Add explicit route for /providers/index.txt to redirect to /providers
-    @app.get("/providers/index.txt", include_in_schema=False)
-    async def redirect_providers_index_txt() -> RedirectResponse:
-        return RedirectResponse("/providers")
-
-    @app.get("/settings", include_in_schema=False)
-    async def serve_settings_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "settings" / "index.html")
-
-    # Add explicit route for /settings/index.txt to redirect to /settings
-    @app.get("/settings/index.txt", include_in_schema=False)
-    async def redirect_settings_index_txt() -> RedirectResponse:
-        return RedirectResponse("/settings")
-
-    @app.get("/transactions", include_in_schema=False)
-    async def serve_transactions_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "transactions" / "index.html")
-
-    # Add explicit route for /transactions/index.txt to redirect to /transactions
-    @app.get("/transactions/index.txt", include_in_schema=False)
-    async def redirect_transactions_index_txt() -> RedirectResponse:
-        return RedirectResponse("/transactions")
-
-    @app.get("/balances", include_in_schema=False)
-    async def serve_balances_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "balances" / "index.html")
-
-    # Add explicit route for /balances/index.txt to redirect to /balances
-    @app.get("/balances/index.txt", include_in_schema=False)
-    async def redirect_balances_index_txt() -> RedirectResponse:
-        return RedirectResponse("/balances")
-
-    @app.get("/logs", include_in_schema=False)
-    async def serve_logs_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "logs" / "index.html")
-
-    # Add explicit route for /logs/index.txt to redirect to /logs
-    @app.get("/logs/index.txt", include_in_schema=False)
-    async def redirect_logs_index_txt() -> RedirectResponse:
-        return RedirectResponse("/logs")
-
-    @app.get("/usage", include_in_schema=False)
-    async def serve_usage_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "usage" / "index.html")
-
-    # Add explicit route for /usage/index.txt to redirect to /usage
-    @app.get("/usage/index.txt", include_in_schema=False)
-    async def redirect_usage_index_txt() -> RedirectResponse:
-        return RedirectResponse("/usage")
-
-    @app.get("/unauthorized", include_in_schema=False)
-    async def serve_unauthorized_ui() -> FileResponse:
-        return FileResponse(UI_DIST_PATH / "unauthorized" / "index.html")
-
-    # Add explicit route for /unauthorized/index.txt to redirect to /unauthorized
-    @app.get("/unauthorized/index.txt", include_in_schema=False)
-    async def redirect_unauthorized_index_txt() -> RedirectResponse:
-        return RedirectResponse("/unauthorized")
-
     @app.get("/favicon.ico", include_in_schema=False)
     async def serve_favicon() -> FileResponse:
         icon_path = UI_DIST_PATH / "icon.ico"
@@ -324,12 +390,12 @@ if UI_DIST_PATH.exists() and UI_DIST_PATH.is_dir():
     async def serve_icon() -> FileResponse:
         return FileResponse(UI_DIST_PATH / "icon.ico")
 
-    app.mount(
-        "/static", StaticFiles(directory=UI_DIST_PATH, check_dir=True), name="ui-static"
-    )
 else:
     logger.warning(
-        f"UI dist directory not found at {UI_DIST_PATH}, skipping static file serving"
+        "UI dist directory not found at %s; serving API only. Run `make ui-build` "
+        "to build the static UI served from here, or `make ui-dev` for the Next.js "
+        "dev server with hot reload on :3000 (it targets this backend on :8000).",
+        UI_DIST_PATH,
     )
 
     @app.get("/", include_in_schema=False)
@@ -346,6 +412,7 @@ else:
 app.include_router(models_router)
 app.include_router(admin_router)
 app.include_router(balance_router)
+app.include_router(lightning_router)
 app.include_router(deprecated_wallet_router)
 app.include_router(providers_router)
 app.include_router(proxy_router)

routstr/core/middleware.py

@@ -14,8 +14,54 @@ logger = get_logger(__name__)
 request_id_context: ContextVar[str | None] = ContextVar("request_id")
 
 
+# Methods that are never logged: HEAD requests are health probes from
+# monitoring/load balancers, OPTIONS are CORS preflights — both are framework
+# chatter, not user-meaningful events.
+_SKIP_LOG_METHODS: frozenset[str] = frozenset({"HEAD", "OPTIONS"})
+
+# Path prefixes to skip. Includes Next.js static chunks and the admin
+# dashboard's internal polling API (/admin/api/*) which the UI hits on a timer
+# to refresh balances, logs, providers, etc. — high volume, low diagnostic
+# value. Mutating admin actions are recorded separately in the audit log.
+_SKIP_LOG_PREFIXES: tuple[str, ...] = (
+    "/_next/",
+    "/admin/api/",
+)
+
+# Exact paths to skip. RSC payload prefetches (`*/index.txt`) fire automatically
+# as the user hovers near `<Link>`s, and `/v1/wallet/info` is polled by the UI.
+_SKIP_LOG_EXACT: frozenset[str] = frozenset(
+    {
+        "/favicon.ico",
+        "/icon.ico",
+        "/v1/wallet/info",
+        "/index.txt",
+        "/login/index.txt",
+        "/model/index.txt",
+        "/providers/index.txt",
+        "/settings/index.txt",
+        "/transactions/index.txt",
+        "/balances/index.txt",
+        "/logs/index.txt",
+        "/usage/index.txt",
+        "/unauthorized/index.txt",
+    }
+)
+
+
+def _should_log(method: str, path: str) -> bool:
+    if method in _SKIP_LOG_METHODS:
+        return False
+    if path in _SKIP_LOG_EXACT:
+        return False
+    return not any(path.startswith(prefix) for prefix in _SKIP_LOG_PREFIXES)
+
+
 class LoggingMiddleware(BaseHTTPMiddleware):
-    """Middleware to log detailed request and response information."""
+    """Middleware to log proxy interactions and page navigation.
+
+    Skips logging for static assets and Next.js chunks to avoid noise.
+    """
 
     async def dispatch(self, request: Request, call_next: Callable) -> Response:
         # Generate request ID
@@ -25,62 +71,20 @@ class LoggingMiddleware(BaseHTTPMiddleware):
         # Set request ID in context for logging
         token = request_id_context.set(request_id)
 
+        path = request.url.path
+        should_log = _should_log(request.method, path)
+
         # Start timing
         start_time = time.time()
 
-        # Log request details
-        request_body = None
-        if request.method in ["POST", "PUT", "PATCH"]:
-            try:
-                # Only read body for non-streaming requests
-                if hasattr(request, "_body"):
-                    request_body = await request.body()
-            except Exception:
-                pass
-
-        # Extract request info
-        client_host = None
-        if request.client:
-            client_host = request.client.host
-
-        # Log incoming request
-        logger.info(
-            "Incoming request",
-            extra={
-                "request_id": request_id,
-                "method": request.method,
-                "path": request.url.path,
-                "query_params": dict(request.query_params),
-                "client_host": client_host,
-                "headers": {
-                    k: v
-                    for k, v in request.headers.items()
-                    if k.lower()
-                    not in [
-                        "authorization",
-                        "x-cashu",
-                        "cookie",
-                        "cf-connecting-ip",
-                        "cf-ipcountry",
-                        "x-forwarded-for",
-                        "x-real-ip",
-                    ]
-                },
-                "body_size": len(request_body) if request_body else 0,
-            },
-        )
-
-        # Log at TRACE level for full body (security filter will redact sensitive data)
-        if request_body and hasattr(logger, "exception"):
-            logger.exception(
-                "Request body",
+        if should_log:
+            logger.info(
+                "Incoming request",
                 extra={
                     "request_id": request_id,
                     "method": request.method,
-                    "path": request.url.path,
-                    "body": request_body.decode("utf-8", errors="ignore")[
-                        :1000
-                    ],  # Limit size
+                    "path": path,
+                    "query_params": dict(request.query_params),
                 },
             )
 
@@ -88,39 +92,33 @@ class LoggingMiddleware(BaseHTTPMiddleware):
         try:
             response = await call_next(request)
 
-            # Calculate duration
-            duration = time.time() - start_time
-
-            # Log response
-            logger.info(
-                "Request completed",
-                extra={
-                    "request_id": request_id,
-                    "method": request.method,
-                    "path": request.url.path,
-                    "status_code": response.status_code,
-                    "duration_ms": round(duration * 1000, 2),
-                    "client_host": client_host,
-                },
-            )
+            if should_log:
+                duration = time.time() - start_time
+                logger.info(
+                    "Request completed",
+                    extra={
+                        "request_id": request_id,
+                        "method": request.method,
+                        "path": path,
+                        "status_code": response.status_code,
+                        "duration_ms": round(duration * 1000, 2),
+                    },
+                )
             if hasattr(response, "headers"):
                 response.headers["x-routstr-request-id"] = request_id
 
             return response
 
         except Exception as e:
-            # Calculate duration
+            # Always log failures, even for skipped paths, so we don't lose errors.
             duration = time.time() - start_time
-
-            # Log error
             logger.error(
                 "Request failed",
                 extra={
                     "request_id": request_id,
                     "method": request.method,
-                    "path": request.url.path,
+                    "path": path,
                     "duration_ms": round(duration * 1000, 2),
-                    "client_host": client_host,
                     "error": str(e),
                     "error_type": type(e).__name__,
                 },

routstr/core/not_found.py

@@ -0,0 +1,55 @@
+"""Shared 404 handler used by the proxy catch-all and tests."""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+from fastapi import Request
+from fastapi.responses import HTMLResponse, JSONResponse, Response
+
+_NOT_FOUND_HTML_FILE = Path(__file__).parent.parent.parent / "ui_out" / "404.html"
+
+
+def _read_not_found_html() -> str | None:
+    try:
+        return _NOT_FOUND_HTML_FILE.read_text(encoding="utf-8")
+    except OSError:
+        return None
+
+
+_NOT_FOUND_HTML: str | None = _read_not_found_html()
+
+
+def build_not_found_response(request: Request, path: str) -> Response:
+    """Return a 404 response.
+
+    HTML 404 page only for GET requests from browsers (Accept: text/html).
+    All POST requests and API clients receive a JSON 404.
+    """
+    accept = request.headers.get("accept", "").lower()
+    prefers_html = (
+        request.method == "GET"
+        and "text/html" in accept
+        and "application/json" not in accept
+    )
+    request_id = getattr(request.state, "request_id", "unknown")
+
+    if prefers_html and _NOT_FOUND_HTML is not None:
+        return HTMLResponse(content=_NOT_FOUND_HTML, status_code=404)
+
+    return JSONResponse(
+        status_code=404,
+        content={
+            "error": {
+                "message": f"Path '/{path}' not found",
+                "type": "not_found",
+                "code": 404,
+            },
+            "request_id": request_id,
+        },
+    )
+
+
+async def not_found_catch_all(request: Request, path: str) -> Response:
+    """ASGI handler form of :func:`build_not_found_response`."""
+    return build_not_found_response(request, path)

routstr/core/provider_slugs.py

@@ -0,0 +1,65 @@
+from __future__ import annotations
+
+import re
+from itertools import count
+from typing import Collection
+
+from sqlmodel import select
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from .db import UpstreamProviderRow
+
+_SLUG_BASE_PATTERN = re.compile(r"[^a-z0-9]+")
+_MAX_SLUG_LENGTH = 64
+
+
+def provider_slug_base(provider_type: str) -> str:
+    """Return a deterministic slug base for a provider type."""
+    base = _SLUG_BASE_PATTERN.sub("-", provider_type.lower()).strip("-")
+    if not base:
+        base = "provider"
+    elif base.isdigit():
+        base = f"provider-{base}"
+    elif len(base) < 3:
+        base = f"{base}-provider"
+
+    if len(base) > _MAX_SLUG_LENGTH:
+        base = base[:_MAX_SLUG_LENGTH].rstrip("-") or "provider"
+    return base
+
+
+def provider_slug_candidate(base: str, suffix_number: int) -> str:
+    if suffix_number == 1:
+        return base
+
+    suffix = f"-{suffix_number}"
+    max_base_length = _MAX_SLUG_LENGTH - len(suffix)
+    return f"{base[:max_base_length].rstrip('-')}{suffix}"
+
+
+async def allocate_unique_provider_slug(
+    session: AsyncSession,
+    provider_type: str,
+    reserved_slugs: Collection[str] = (),
+) -> str:
+    """Allocate a stable, deterministic provider slug.
+
+    The first provider of a type gets ``openai``; later collisions get
+    ``openai-2``, ``openai-3``, etc. ``reserved_slugs`` covers rows staged in
+    memory but not flushed yet, such as settings/env seeding.
+    """
+    base = provider_slug_base(provider_type)
+    reserved = {slug.lower() for slug in reserved_slugs}
+
+    for suffix_number in count(1):
+        candidate = provider_slug_candidate(base, suffix_number)
+        if candidate in reserved:
+            continue
+
+        result = await session.exec(
+            select(UpstreamProviderRow).where(UpstreamProviderRow.slug == candidate)
+        )
+        if result.first() is None:
+            return candidate
+
+    raise RuntimeError("unreachable")

routstr/core/redaction.py

@@ -0,0 +1,51 @@
+"""Redaction helpers for sensitive provider identifiers.
+
+Single source of truth for stripping account-scoped identifiers (e.g. OpenAI
+organization IDs) from any text before it is logged, returned to a caller, or
+written to an audit entry.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Any
+
+# OpenAI-style organization identifiers look like ``org-<base62>``. Require at
+# least 6 trailing chars so the already-redacted literal ``org-[REDACTED]`` is
+# never re-matched (``[`` is not in the character class).
+_ORG_ID_PATTERN = re.compile(r"\borg-[A-Za-z0-9]{6,}\b")
+
+ORG_ID_PLACEHOLDER = "org-[REDACTED]"
+
+
+def redact_org_ids(text: str) -> str:
+    """Replace OpenAI-style organization IDs with ``org-[REDACTED]``.
+
+    Args:
+        text: Arbitrary text that may embed an ``org-*`` identifier.
+
+    Returns:
+        The text with every organization ID replaced. Non-string input is
+        returned unchanged after coercion to ``str``.
+    """
+    if not text:
+        return text
+    return _ORG_ID_PATTERN.sub(ORG_ID_PLACEHOLDER, text)
+
+
+def redact_obj(obj: Any) -> Any:
+    """Recursively redact organization IDs in arbitrary nested structures.
+
+    Strings are redacted in place; dicts and lists/tuples are walked so that
+    identifiers nested inside structured payloads (e.g. log ``extra`` fields or
+    error ``details``) are also stripped. Other types are returned unchanged.
+    """
+    if isinstance(obj, str):
+        return redact_org_ids(obj)
+    if isinstance(obj, dict):
+        return {key: redact_obj(value) for key, value in obj.items()}
+    if isinstance(obj, list):
+        return [redact_obj(value) for value in obj]
+    if isinstance(obj, tuple):
+        return tuple(redact_obj(value) for value in obj)
+    return obj

routstr/core/settings.py

@@ -41,6 +41,15 @@ class Settings(BaseSettings):
     primary_mint: str = Field(default="", env="PRIMARY_MINT_URL")
     primary_mint_unit: str = Field(default="sat", env="PRIMARY_MINT_UNIT")
 
+    # Lightning payout configuration
+    # Minimum available balance (in satoshis) before profit is paid out over
+    # Lightning
+    min_payout_sat: int = Field(default=210, gt=0, env="MIN_PAYOUT_SAT")
+    # Interval (seconds) between periodic payout attempts. Must be positive.
+    payout_interval_seconds: int = Field(
+        default=900, gt=0, env="PAYOUT_INTERVAL_SECONDS"
+    )
+
     # Pricing
     # Default behavior: derive pricing from MODELS
     # If fixed_pricing is True -> use fixed_cost_per_request and ignore tokens
@@ -52,12 +61,26 @@ class Settings(BaseSettings):
     exchange_fee: float = Field(default=1.005, env="EXCHANGE_FEE")
     upstream_provider_fee: float = Field(default=1.05, env="UPSTREAM_PROVIDER_FEE")
     tolerance_percentage: float = Field(default=1.0, env="TOLERANCE_PERCENTAGE")
-    child_key_cost: int = Field(default=1000, env="CHILD_KEY_COST")
+    child_key_cost: int = Field(default=0, env="CHILD_KEY_COST")
     # Minimum per-request charge in millisatoshis when model pricing is free/zero
     min_request_msat: int = Field(default=1, env="MIN_REQUEST_MSAT")
     reset_reserved_balance_on_startup: bool = Field(
         default=True, env="RESET_RESERVED_BALANCE_ON_STARTUP"
     )  # deactivate in horizontal scaling setups
+    # Reservations older than this are considered leaked (client disconnect,
+    # crash, abandoned stream) and released by the background sweeper and the
+    # refund endpoint.
+    stale_reservation_timeout_seconds: int = Field(
+        default=300, env="STALE_RESERVATION_TIMEOUT_SECONDS"
+    )
+    # Background prune of dead (zero balance, never used) API keys.
+    # Interval 0 disables it; min-age is a grace period (default 1 week).
+    dead_key_prune_interval_seconds: int = Field(
+        default=3600, env="DEAD_KEY_PRUNE_INTERVAL_SECONDS"
+    )
+    dead_key_min_age_seconds: int = Field(
+        default=604_800, env="DEAD_KEY_MIN_AGE_SECONDS"
+    )
 
     # Network
     cors_origins: list[str] = Field(default_factory=lambda: ["*"], env="CORS_ORIGINS")
@@ -74,6 +97,7 @@ class Settings(BaseSettings):
     enable_pricing_refresh: bool = Field(default=True, env="ENABLE_PRICING_REFRESH")
     enable_models_refresh: bool = Field(default=True, env="ENABLE_MODELS_REFRESH")
     refund_cache_ttl_seconds: int = Field(default=3600, env="REFUND_CACHE_TTL_SECONDS")
+    refund_sweep_ttl_seconds: int = Field(default=604800, env="REFUND_SWEEP_TTL_SECONDS")
 
     # Logging
     log_level: str = Field(default="INFO", env="LOG_LEVEL")
@@ -92,6 +116,20 @@ class Settings(BaseSettings):
 
     # Discovery
     relays: list[str] = Field(default_factory=list, env="RELAYS")
+    enable_analytics_sharing: bool = Field(
+        default=True, env="ENABLE_ANALYTICS_SHARING"
+    )
+
+def _normalize_settings_data(data: dict[str, Any]) -> dict[str, Any]:
+    """Discard unknown keys from persisted settings."""
+    normalized: dict[str, Any] = {}
+    known_fields = Settings.__fields__
+
+    for key, value in data.items():
+        if key in known_fields:
+            normalized[key] = value
+
+    return normalized
 
 
 def _compute_primary_mint(cashu_mints: list[str]) -> str:
@@ -231,16 +269,21 @@ class SettingsService:
 
             db_id, db_data, _updated_at = row
             try:
-                db_json = (
+                db_json_raw = (
                     json.loads(db_data) if isinstance(db_data, str) else dict(db_data)
                 )
+                if not isinstance(db_json_raw, dict):
+                    db_json_raw = {}
             except Exception:
-                db_json = {}
+                db_json_raw = {}
+            db_json = _normalize_settings_data(db_json_raw)
 
+            valid_fields = set(env_resolved.dict().keys())
             merged_dict: dict[str, Any] = dict(env_resolved.dict())
             merged_dict.update(
-                {k: v for k, v in db_json.items() if v not in (None, "", [], {})}
+                {k: v for k, v in db_json.items() if v not in (None, "", [], {}) and k in valid_fields}
             )
+            merged_dict = Settings(**merged_dict).dict()
 
             # Ensure primary_mint is consistent with cashu_mints if not explicitly set
             if not merged_dict.get("primary_mint"):
@@ -248,7 +291,7 @@ class SettingsService:
                     merged_dict.get("cashu_mints", [])
                 )
 
-            if any(k not in db_json for k in merged_dict.keys()):
+            if db_json_raw != merged_dict:
                 await db_session.exec(  # type: ignore
                     text(
                         "UPDATE settings SET data = :data, updated_at = :updated_at WHERE id = 1"
@@ -271,7 +314,7 @@ class SettingsService:
     ) -> Settings:
         async with cls._lock:
             current = cls.get()
-            candidate_dict = {**current.dict(), **partial}
+            candidate_dict = {**current.dict(), **_normalize_settings_data(partial)}
             candidate = Settings(**candidate_dict)
             from sqlmodel import text
 
@@ -305,8 +348,10 @@ class SettingsService:
                 raise RuntimeError("Settings row missing")
             (data_str,) = row
             data = json.loads(data_str) if isinstance(data_str, str) else dict(data_str)
+            valid_fields = set(settings.dict().keys())
             # Update in-place
             for k, v in data.items():
-                setattr(settings, k, v)
+                if k in valid_fields:
+                    setattr(settings, k, v)
             cls._current = settings
             return settings

routstr/core/version.py

@@ -0,0 +1,81 @@
+"""Application version resolution.
+
+Priority order:
+
+1. ``VERSION_SUFFIX`` env var (manual override; preserves prior behaviour).
+2. Bare base version when HEAD is on the matching release tag (detected via
+   ``GIT_TAG`` env or ``git describe --tags --exact-match HEAD``).
+3. ``GIT_COMMIT`` env var (build-time injection) -> ``<base>+g<sha>``.
+4. Local ``.git`` lookup (source checkouts) -> ``<base>+g<sha>``.
+5. Fallback: bare base version.
+
+The ``+g<sha>`` form is PEP 440 local-version syntax so the result remains a
+valid package version.
+"""
+
+from __future__ import annotations
+
+import os
+import subprocess
+from functools import lru_cache
+from pathlib import Path
+
+BASE_VERSION = "0.4.4"
+
+_REPO_ROOT = Path(__file__).resolve().parents[2]
+_GIT_TIMEOUT_SECONDS = 2.0
+
+
+def _run_git(*args: str) -> str | None:
+    try:
+        result = subprocess.run(  # noqa: S603 - fixed argv, no shell
+            ["git", *args],
+            cwd=_REPO_ROOT,
+            check=False,
+            capture_output=True,
+            text=True,
+            timeout=_GIT_TIMEOUT_SECONDS,
+        )
+    except (FileNotFoundError, subprocess.SubprocessError, OSError):
+        return None
+    if result.returncode != 0:
+        return None
+    return result.stdout.strip() or None
+
+
+def _git_short_sha() -> str | None:
+    sha = os.getenv("GIT_COMMIT", "").strip()
+    if sha:
+        return sha[:7]
+    return _run_git("rev-parse", "--short=7", "HEAD")
+
+
+def _on_tagged_release() -> bool:
+    tag = os.getenv("GIT_TAG", "").strip()
+    if tag:
+        return tag.lstrip("v") == BASE_VERSION
+    described = _run_git("describe", "--tags", "--exact-match", "HEAD")
+    if described and described.lstrip("v") == BASE_VERSION:
+        return True
+    return False
+
+
+@lru_cache(maxsize=1)
+def get_version() -> str:
+    suffix = os.getenv("VERSION_SUFFIX")
+    if suffix is not None:
+        return f"{BASE_VERSION}-{suffix}"
+
+    if _on_tagged_release():
+        return BASE_VERSION
+
+    sha = _git_short_sha()
+    if not sha:
+        return BASE_VERSION
+
+    return f"{BASE_VERSION}+g{sha}"
+
+
+__version__ = get_version()
+
+__all__ = ["BASE_VERSION", "__version__", "get_version"]

routstr/lightning.py

@@ -1,13 +1,14 @@
+import asyncio
 import hashlib
 import secrets
 import time
 
-from fastapi import APIRouter, Depends, HTTPException
+from fastapi import APIRouter, Depends, Header, HTTPException
 from pydantic import BaseModel, Field
-from sqlmodel import select
+from sqlmodel import col, select
 from sqlmodel.ext.asyncio.session import AsyncSession
 
-from .core.db import ApiKey, LightningInvoice, get_session
+from .core.db import ApiKey, LightningInvoice, create_session, get_session
 from .core.logging import get_logger
 from .core.settings import settings
 from .wallet import get_wallet
@@ -19,15 +20,29 @@ lightning_router = APIRouter(prefix="/lightning")
 
 class InvoiceCreateRequest(BaseModel):
     amount_sats: int = Field(gt=0, le=1_000_000, description="Amount in satoshis")
-    purpose: str = Field(description="create or topup", pattern="^(create|topup)$")
+    purpose: str = Field(
+        default="create",
+        description="create or topup",
+        pattern="^(create|topup)$",
+    )
     api_key: str | None = Field(
-        default=None, description="Required for topup operations"
+        default=None,
+        description="Deprecated: legacy field for topup. Prefer Authorization header.",
     )
     balance_limit: int | None = Field(default=None)
     balance_limit_reset: str | None = Field(default=None)
     validity_date: int | None = Field(default=None)
 
 
+def _extract_bearer_api_key(authorization: str | None) -> str | None:
+    if not authorization:
+        return None
+    token = authorization.strip()
+    if token.lower().startswith("bearer "):
+        token = token[7:].strip()
+    return token or None
+
+
 class InvoiceCreateResponse(BaseModel):
     invoice_id: str
     bolt11: str
@@ -64,18 +79,21 @@ def generate_invoice_id() -> str:
 @lightning_router.post("/invoice", response_model=InvoiceCreateResponse)
 async def create_invoice(
     request: InvoiceCreateRequest,
+    authorization: str | None = Header(default=None),
     session: AsyncSession = Depends(get_session),
 ) -> InvoiceCreateResponse:
-    if request.purpose == "topup" and not request.api_key:
-        raise HTTPException(
-            status_code=400, detail="api_key is required for topup operations"
-        )
+    api_key_token = _extract_bearer_api_key(authorization) or request.api_key
 
-    if request.purpose == "topup" and request.api_key:
-        if not request.api_key.startswith("sk-"):
+    if request.purpose == "topup":
+        if not api_key_token:
+            raise HTTPException(
+                status_code=401,
+                detail="Authorization bearer api key is required for topup",
+            )
+        if not api_key_token.startswith("sk-"):
             raise HTTPException(status_code=400, detail="Invalid API key format")
 
-        api_key = await session.get(ApiKey, request.api_key[3:])
+        api_key = await session.get(ApiKey, api_key_token[3:])
         if not api_key:
             raise HTTPException(status_code=404, detail="API key not found")
 
@@ -95,7 +113,7 @@ async def create_invoice(
             description=description,
             payment_hash=payment_hash,
             status="pending",
-            api_key_hash=request.api_key[3:] if request.api_key else None,
+            api_key_hash=api_key_token[3:] if api_key_token else None,
             purpose=request.purpose,
             balance_limit=request.balance_limit,
             balance_limit_reset=request.balance_limit_reset,
@@ -142,13 +160,13 @@ async def get_invoice_status(
     if not invoice:
         raise HTTPException(status_code=404, detail="Invoice not found")
 
+    if invoice.status == "pending":
+        await check_invoice_payment(invoice, session)
+
     if invoice.status == "pending" and int(time.time()) > invoice.expires_at:
         invoice.status = "expired"
         await session.commit()
 
-    if invoice.status == "pending":
-        await check_invoice_payment(invoice, session)
-
     api_key = None
     if invoice.status == "paid" and invoice.purpose == "create":
         if invoice.api_key_hash:
@@ -251,6 +269,9 @@ async def create_api_key_from_invoice(
         balance=invoice.amount_sats * 1000,  # Convert to msats
         refund_currency="sat",
         refund_mint_url=settings.primary_mint,
+        balance_limit=invoice.balance_limit,
+        balance_limit_reset=invoice.balance_limit_reset,
+        validity_date=invoice.validity_date,
     )
 
     session.add(api_key)
@@ -274,3 +295,41 @@ async def topup_api_key_from_invoice(
 
     api_key.balance += invoice.amount_sats * 1000  # Convert to msats
     await session.flush()
+
+
+INVOICE_WATCH_INTERVAL_SECONDS = 5
+INVOICE_WATCH_BATCH_LIMIT = 100
+
+
+async def periodic_invoice_watcher() -> None:
+    """Background task: detect paid Lightning invoices and credit balances.
+
+    Removes the need for clients to poll the status endpoint after paying.
+    """
+    while True:
+        try:
+            async with create_session() as session:
+                now = int(time.time())
+                result = await session.exec(
+                    select(LightningInvoice)
+                    .where(
+                        LightningInvoice.status == "pending",
+                        col(LightningInvoice.expires_at) > now,
+                    )
+                    .limit(INVOICE_WATCH_BATCH_LIMIT)
+                )
+                pending = result.all()
+                for invoice in pending:
+                    try:
+                        await check_invoice_payment(invoice, session)
+                    except Exception as e:
+                        logger.error(
+                            "Invoice watcher failed for invoice",
+                            extra={"invoice_id": invoice.id, "error": str(e)},
+                        )
+        except asyncio.CancelledError:
+            raise
+        except Exception as e:
+            logger.error(f"Invoice watcher loop error: {e}")
+
+        await asyncio.sleep(INVOICE_WATCH_INTERVAL_SECONDS)

routstr/nostr/__init__.py

@@ -1,4 +1,5 @@
+from .analytics import publish_usage_analytics
 from .discovery import providers_cache_refresher
 from .listing import announce_provider
 
-__all__ = ["providers_cache_refresher", "announce_provider"]
+__all__ = ["providers_cache_refresher", "announce_provider", "publish_usage_analytics"]

routstr/nostr/analytics.py

@@ -0,0 +1,419 @@
+#!/usr/bin/env python3
+"""
+Nostr usage analytics publisher.
+Publishes a single replaceable analytics snapshot for each provider.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import hashlib
+import json
+import time
+from typing import Any
+
+from nostr.event import Event
+from nostr.key import PrivateKey
+
+from ..core import get_logger
+from ..core.log_manager import log_manager
+from ..core.settings import settings
+from .listing import nsec_to_keypair, publish_to_relay
+
+logger = get_logger(__name__)
+
+ANALYTICS_KIND = 38422
+ANALYTICS_SCHEMA = "routstr.analytics.snapshot.v1"
+DEFAULT_RELAYS = [
+    "wss://relay.nostr.band",
+    "wss://relay.damus.io",
+    "wss://relay.routstr.com",
+    "wss://nos.lol",
+]
+PUBLISH_INTERVAL_SECONDS = 15 * 60
+DISABLED_POLL_SECONDS = 60
+DASHBOARD_WINDOW_HOURS = 24
+DASHBOARD_INTERVAL_MINUTES = 60
+MODEL_LIMIT = 20
+WINDOW_DEFINITIONS: tuple[tuple[str, int, int], ...] = (
+    ("24h", 24, 60),
+    ("7d", 7 * 24, 6 * 60),
+    ("30d", 30 * 24, 24 * 60),
+    ("3m", 90 * 24, 24 * 60),
+    ("1y", 365 * 24, 7 * 24 * 60),
+)
+
+
+def _event_to_dict(ev: Event) -> dict[str, Any]:
+    return {
+        "id": ev.id,
+        "pubkey": ev.public_key,
+        "created_at": ev.created_at,
+        "kind": int(ev.kind) if not isinstance(ev.kind, int) else ev.kind,
+        "tags": ev.tags,
+        "content": ev.content,
+        "sig": ev.signature,
+    }
+
+
+def _resolve_provider_id(public_key_hex: str) -> str:
+    explicit_provider_id = (settings.provider_id or "").strip()
+    if explicit_provider_id:
+        return explicit_provider_id
+    return public_key_hex[:12]
+
+
+def _resolve_endpoint_urls() -> list[str]:
+    urls: list[str] = []
+    http_url = (settings.http_url or "").strip()
+    onion_url = (settings.onion_url or "").strip()
+
+    if http_url and http_url != "http://localhost:8000":
+        urls.append(http_url)
+
+    if onion_url:
+        if onion_url.endswith(".onion") and not (
+            onion_url.startswith("http://") or onion_url.startswith("https://")
+        ):
+            onion_url = f"http://{onion_url}"
+        urls.append(onion_url)
+
+    return urls
+
+
+def _resolve_relays() -> list[str]:
+    configured = [url.strip() for url in settings.relays if url.strip()]
+    return configured if configured else list(DEFAULT_RELAYS)
+
+
+def _to_int(value: Any) -> int:
+    if isinstance(value, bool):
+        return int(value)
+    if isinstance(value, int):
+        return value
+    if isinstance(value, float):
+        return int(value)
+    if isinstance(value, str):
+        try:
+            return int(float(value))
+        except ValueError:
+            return 0
+    return 0
+
+
+def _to_float(value: Any) -> float:
+    if isinstance(value, bool):
+        return float(int(value))
+    if isinstance(value, (int, float)):
+        return float(value)
+    if isinstance(value, str):
+        try:
+            return float(value)
+        except ValueError:
+            return 0.0
+    return 0.0
+
+
+def _aggregate_top_model_usage(
+    model_usage_mix: dict[str, Any],
+) -> tuple[list[dict[str, Any]], dict[str, Any]]:
+    top_models_raw = model_usage_mix.get("top_models", [])
+    mix_metrics_raw = model_usage_mix.get("metrics", [])
+
+    top_models = [model for model in top_models_raw if isinstance(model, str)]
+    metrics = [row for row in mix_metrics_raw if isinstance(row, dict)]
+
+    model_totals: dict[str, dict[str, float | int]] = {
+        model: {
+            "successful_requests": 0,
+            "revenue_msats": 0.0,
+            "total_tokens": 0,
+        }
+        for model in top_models
+    }
+    others = {
+        "successful_requests": 0,
+        "revenue_msats": 0.0,
+        "total_tokens": 0,
+    }
+
+    for metric in metrics:
+        model_counts = metric.get("model_counts", {})
+        model_revenue = metric.get("model_revenue_msats", {})
+        model_tokens = metric.get("model_tokens", {})
+
+        if isinstance(model_counts, dict):
+            for model, count in model_counts.items():
+                if model in model_totals:
+                    model_totals[model]["successful_requests"] += _to_int(count)
+
+        if isinstance(model_revenue, dict):
+            for model, amount in model_revenue.items():
+                if model in model_totals:
+                    model_totals[model]["revenue_msats"] += _to_float(amount)
+
+        if isinstance(model_tokens, dict):
+            for model, token_count in model_tokens.items():
+                if model in model_totals:
+                    model_totals[model]["total_tokens"] += _to_int(token_count)
+
+        others["successful_requests"] += _to_int(metric.get("others", 0))
+        others["revenue_msats"] += _to_float(metric.get("others_revenue_msats", 0.0))
+        others["total_tokens"] += _to_int(metric.get("others_tokens", 0))
+
+    model_rows = [
+        {
+            "model": model,
+            "successful_requests": int(values["successful_requests"]),
+            "revenue_msats": float(values["revenue_msats"]),
+            "total_tokens": int(values["total_tokens"]),
+        }
+        for model, values in model_totals.items()
+    ]
+    model_rows.sort(
+        key=lambda row: _to_int(row.get("successful_requests", 0)),
+        reverse=True,
+    )
+
+    return model_rows, others
+
+
+def _build_summary_payload(summary: dict[str, Any]) -> dict[str, Any]:
+    return {
+        "total_requests": _to_int(summary.get("total_requests", 0)),
+        "successful_chat_completions": _to_int(
+            summary.get("successful_chat_completions", 0)
+        ),
+        "failed_requests": _to_int(summary.get("failed_requests", 0)),
+        "success_rate": _to_float(summary.get("success_rate", 0.0)),
+        "unique_models_count": _to_int(summary.get("unique_models_count", 0)),
+        "input_tokens": _to_int(summary.get("input_tokens", 0)),
+        "output_tokens": _to_int(summary.get("output_tokens", 0)),
+        "total_tokens": _to_int(summary.get("total_tokens", 0)),
+        "revenue_msats": _to_float(summary.get("revenue_msats", 0.0)),
+        "refunds_msats": _to_float(summary.get("refunds_msats", 0.0)),
+        "net_revenue_msats": _to_float(summary.get("net_revenue_msats", 0.0)),
+        "revenue_sats": _to_float(summary.get("revenue_sats", 0.0)),
+        "refunds_sats": _to_float(summary.get("refunds_sats", 0.0)),
+        "net_revenue_sats": _to_float(summary.get("net_revenue_sats", 0.0)),
+    }
+
+
+def _build_window_payload(
+    *,
+    hours: int,
+    interval_minutes: int,
+    model_limit: int,
+) -> dict[str, Any]:
+    dashboard = log_manager.get_usage_dashboard(
+        interval=interval_minutes,
+        hours=hours,
+        error_limit=1,
+        model_limit=model_limit,
+    )
+
+    summary = dashboard.get("summary", {})
+    model_usage_mix = dashboard.get("model_usage_mix", {})
+
+    summary_payload = _build_summary_payload(summary if isinstance(summary, dict) else {})
+    usage_mix_payload = model_usage_mix if isinstance(model_usage_mix, dict) else {}
+    top_model_usage, others_usage = _aggregate_top_model_usage(usage_mix_payload)
+
+    return {
+        "window_hours": hours,
+        "interval_minutes": interval_minutes,
+        "summary": summary_payload,
+        "model_usage_mix": usage_mix_payload,
+        "top_model_usage": top_model_usage,
+        "others_usage": others_usage,
+    }
+
+
+def build_stats_snapshot_payload(
+    provider_id: str,
+    *,
+    public_key_hex: str,
+    generated_at: int,
+    window_hours: int = DASHBOARD_WINDOW_HOURS,
+    interval_minutes: int = DASHBOARD_INTERVAL_MINUTES,
+    model_limit: int = MODEL_LIMIT,
+) -> dict[str, Any]:
+    _ = (window_hours, interval_minutes)
+    windows: dict[str, dict[str, Any]] = {}
+    for key, hours, window_interval_minutes in WINDOW_DEFINITIONS:
+        windows[key] = _build_window_payload(
+            hours=hours,
+            interval_minutes=window_interval_minutes,
+            model_limit=model_limit,
+        )
+
+    primary_window = windows.get("24h", {})
+    summary_payload = (
+        primary_window.get("summary", {})
+        if isinstance(primary_window.get("summary", {}), dict)
+        else {}
+    )
+    usage_mix_payload = (
+        primary_window.get("model_usage_mix", {})
+        if isinstance(primary_window.get("model_usage_mix", {}), dict)
+        else {}
+    )
+    top_model_usage = (
+        primary_window.get("top_model_usage", [])
+        if isinstance(primary_window.get("top_model_usage", []), list)
+        else []
+    )
+    others_usage = (
+        primary_window.get("others_usage", {})
+        if isinstance(primary_window.get("others_usage", {}), dict)
+        else {}
+    )
+
+    return {
+        "schema": ANALYTICS_SCHEMA,
+        "generated_at": generated_at,
+        "provider_id": provider_id,
+        "pubkey": public_key_hex,
+        "npub": settings.npub or "",
+        "endpoint_urls": _resolve_endpoint_urls(),
+        "window_hours": DASHBOARD_WINDOW_HOURS,
+        "interval_minutes": DASHBOARD_INTERVAL_MINUTES,
+        "summary": summary_payload,
+        "model_usage_mix": usage_mix_payload,
+        "top_model_usage": top_model_usage,
+        "others_usage": others_usage,
+        "windows": windows,
+    }
+
+
+def create_stats_snapshot_event(
+    private_key_hex: str,
+    provider_id: str,
+    payload_json: str,
+    *,
+    d_tag: str,
+) -> dict[str, Any]:
+    private_key = PrivateKey(bytes.fromhex(private_key_hex))
+    tags = [
+        ["d", d_tag],
+        ["provider", provider_id],
+        ["schema", ANALYTICS_SCHEMA],
+    ]
+
+    event = Event(
+        public_key=private_key.public_key.hex(),
+        content=payload_json,
+        kind=ANALYTICS_KIND,
+        tags=tags,
+    )
+    private_key.sign_event(event)
+    return _event_to_dict(event)
+
+
+def _fingerprint_payload(payload: dict[str, Any]) -> str:
+    normalized = dict(payload)
+    # Ignore generated timestamp for semantic dedupe.
+    normalized.pop("generated_at", None)
+    payload_json = json.dumps(normalized, separators=(",", ":"), sort_keys=True)
+    return hashlib.sha256(payload_json.encode("utf-8")).hexdigest()
+
+
+async def publish_usage_analytics() -> None:
+    last_payload_hash: str | None = None
+
+    parsed_nsec: str | None = None
+    private_key_hex: str | None = None
+    public_key_hex: str | None = None
+    provider_id: str | None = None
+    warned_missing_nsec = False
+
+    logger.info("Usage analytics sharing task started")
+
+    while True:
+        try:
+            if not settings.enable_analytics_sharing:
+                await asyncio.sleep(DISABLED_POLL_SECONDS)
+                continue
+
+            nsec = (settings.nsec or "").strip()
+            if not nsec:
+                if not warned_missing_nsec:
+                    logger.info("NSEC is not configured; skipping analytics sharing to Nostr")
+                    warned_missing_nsec = True
+                await asyncio.sleep(DISABLED_POLL_SECONDS)
+                continue
+
+            warned_missing_nsec = False
+            if nsec != parsed_nsec or private_key_hex is None or public_key_hex is None:
+                keypair = nsec_to_keypair(nsec)
+                if not keypair:
+                    logger.error("Invalid NSEC; analytics sharing is paused")
+                    await asyncio.sleep(DISABLED_POLL_SECONDS)
+                    continue
+                private_key_hex, public_key_hex = keypair
+                parsed_nsec = nsec
+                provider_id = _resolve_provider_id(public_key_hex)
+                last_payload_hash = None
+
+            if private_key_hex is None or public_key_hex is None:
+                await asyncio.sleep(DISABLED_POLL_SECONDS)
+                continue
+
+            relay_urls = _resolve_relays()
+            if not relay_urls:
+                logger.warning("No Nostr relays configured; analytics sharing skipped")
+                await asyncio.sleep(DISABLED_POLL_SECONDS)
+                continue
+
+            resolved_provider_id = provider_id or _resolve_provider_id(public_key_hex)
+            now_ts = int(time.time())
+            payload = build_stats_snapshot_payload(
+                resolved_provider_id,
+                public_key_hex=public_key_hex,
+                generated_at=now_ts,
+            )
+
+            payload_hash = _fingerprint_payload(payload)
+            if last_payload_hash == payload_hash:
+                await asyncio.sleep(PUBLISH_INTERVAL_SECONDS)
+                continue
+
+            payload_json = json.dumps(payload, separators=(",", ":"), sort_keys=True)
+            d_tag = f"{resolved_provider_id}:stats"
+            event = create_stats_snapshot_event(
+                private_key_hex,
+                resolved_provider_id,
+                payload_json,
+                d_tag=d_tag,
+            )
+
+            success_count = 0
+            for relay_url in relay_urls:
+                if await publish_to_relay(relay_url, event):
+                    success_count += 1
+
+            if success_count > 0:
+                last_payload_hash = payload_hash
+
+            logger.info(
+                "Published analytics snapshot (success=%s/%s provider=%s)",
+                success_count,
+                len(relay_urls),
+                resolved_provider_id,
+                extra={
+                    "relay_success_count": success_count,
+                    "relay_total": len(relay_urls),
+                    "provider_id": resolved_provider_id,
+                },
+            )
+            await asyncio.sleep(PUBLISH_INTERVAL_SECONDS)
+
+        except asyncio.CancelledError:
+            logger.info("Usage analytics sharing task cancelled")
+            break
+        except Exception as e:
+            logger.error(
+                "Usage analytics sharing error",
+                extra={"error": str(e), "error_type": type(e).__name__},
+            )
+            await asyncio.sleep(DISABLED_POLL_SECONDS)

routstr/nostr/listing.py

@@ -26,7 +26,7 @@ logger = get_logger(__name__)
 
 def get_app_version() -> str | None:
     try:
-        from ..core.main import __version__ as imported_version
+        from ..core.version import __version__ as imported_version
 
         return imported_version
     except Exception:

routstr/payment/cost_calculation.py

@@ -3,9 +3,17 @@ import math
 from pydantic.v1 import BaseModel
 
 from ..core import get_logger
-from ..core.db import AsyncSession
 from ..core.settings import settings
 from .price import sats_usd_price
+from .usage import normalize_usage, parse_token_count
+
+__all__ = [
+    "CostData",
+    "CostDataError",
+    "MaxCostData",
+    "calculate_cost",
+    "parse_token_count",
+]
 
 logger = get_logger(__name__)
 
@@ -16,6 +24,12 @@ class CostData(BaseModel):
     output_msats: int
     total_msats: int
     total_usd: float = 0.0
+    input_tokens: int = 0
+    output_tokens: int = 0
+    cache_read_input_tokens: int = 0
+    cache_creation_input_tokens: int = 0
+    cache_read_msats: int = 0
+    cache_creation_msats: int = 0
 
 
 class MaxCostData(CostData):
@@ -27,11 +41,33 @@ class CostDataError(BaseModel):
     code: str
 
 
-async def calculate_cost(  # todo: can be sync
-    response_data: dict, max_cost: int, session: AsyncSession
-) -> CostData | MaxCostData | CostDataError:
+def _empty_cost(cls: type[CostData] = CostData) -> CostData:
+    """Build an all-zero cost object — a full refund for an empty response.
+
+    Shared by the two paths that must not bill: an upstream response with no
+    usage data at all, and one that reports a USD cost but carries zero tokens
+    in every bucket.
     """
-    Calculate the cost of an API request based on token usage.
+    return cls(
+        base_msats=0,
+        input_msats=0,
+        output_msats=0,
+        total_msats=0,
+        total_usd=0.0,
+        input_tokens=0,
+        output_tokens=0,
+        cache_read_input_tokens=0,
+        cache_creation_input_tokens=0,
+        cache_read_msats=0,
+        cache_creation_msats=0,
+    )
+
+
+async def calculate_cost(
+    response_data: dict,
+    max_cost: int,
+) -> CostData | MaxCostData | CostDataError:
+    """Calculate the cost of an API request based on token usage.
 
     Args:
         response_data: Response data containing usage information
@@ -39,6 +75,9 @@ async def calculate_cost(  # todo: can be sync
 
     Returns:
         Cost data or error information
+
+    The response's usage object is normalized with the default union parser;
+    this function holds no vendor-dialect knowledge of its own.
     """
     logger.debug(
         "Starting cost calculation",
@@ -49,61 +88,95 @@ async def calculate_cost(  # todo: can be sync
         },
     )
 
-    if "usage" not in response_data or response_data["usage"] is None:
+    usage = normalize_usage(response_data.get("usage"))
+
+    if usage is None:
         logger.warning(
-            "No usage data in response, using base cost only",
+            "No usage data in response — billing at MaxCostData with zero "
+            "tokens. Dashboard will show this request as `(0+0)`. Most "
+            "common cause: upstream stream did not include a final usage "
+            "chunk (OpenAI-compat backends require "
+            "`stream_options.include_usage=true`).",
             extra={
                 "max_cost_msats": max_cost,
                 "model": response_data.get("model", "unknown"),
+                "response_keys": sorted(response_data.keys())
+                if isinstance(response_data, dict)
+                else None,
             },
         )
-        return MaxCostData(
-            base_msats=0,
-            input_msats=0,
-            output_msats=0,
-            total_msats=0,
-            total_usd=0.0,
-        )
+        return _empty_cost(MaxCostData)
 
-    usage_data = response_data["usage"]
+    usage_data = response_data.get("usage") or {}
+    if not isinstance(usage_data, dict):
+        usage_data = {}
 
-    usd_cost = 0.0
-
-    # Prioritize cost_details.upstream_inference_cost
-    if "cost_details" in usage_data:
-        usd_cost = float(
-            usage_data["cost_details"].get("upstream_inference_cost", 0) or 0
-        )
-
-    # Fallback to cost field if upstream_inference_cost is 0
-    if usd_cost == 0 and "cost" in usage_data:
-        try:
-            usd_cost = float(usage_data.get("cost", 0) or 0)
-        except Exception:
-            pass
+    input_tokens = usage.input_tokens
+    output_tokens = usage.output_tokens
+    cache_read_tokens = usage.cache_read_tokens
+    cache_creation_tokens = usage.cache_write_tokens
 
+    # Try USD cost first
+    usd_cost = _resolve_usd_cost(usage_data, response_data)
     if usd_cost > 0:
-        try:
-            sats_per_usd = 1.0 / sats_usd_price()
-            cost_in_sats = usd_cost * sats_per_usd
-            cost_in_msats = math.ceil(cost_in_sats * 1000)
-
-            logger.info(
-                "Using cost from usage data/details",
+        truly_empty = (
+            input_tokens == 0
+            and output_tokens == 0
+            and cache_read_tokens == 0
+            and cache_creation_tokens == 0
+        )
+        if truly_empty:
+            logger.warning(
+                "Upstream reported a USD cost but the response carries no "
+                "tokens at all (input, output, cache-read and cache-creation "
+                "are all zero) — refunding in full rather than billing the "
+                "USD-derived cost for an empty response.",
                 extra={
-                    "usd_cost": usd_cost,
-                    "cost_in_sats": cost_in_sats,
-                    "cost_in_msats": cost_in_msats,
                     "model": response_data.get("model", "unknown"),
+                    "usd_cost": usd_cost,
+                    "usage_keys": sorted(usage_data.keys())
+                    if isinstance(usage_data, dict)
+                    else None,
                 },
             )
-
-            return CostData(
-                base_msats=-1,
-                input_msats=-1,  # Cost field doesn't break down by token type
-                output_msats=-1,
-                total_msats=cost_in_msats,
-                total_usd=usd_cost,
+            return _empty_cost()
+        if input_tokens == 0 and output_tokens == 0:
+            logger.warning(
+                "Upstream reported a USD cost but no token counts — "
+                "billing the USD-derived cost while the dashboard will "
+                "show this request as `(0+0)` tokens. Check that the "
+                "upstream actually emits `usage.input_tokens` and "
+                "`usage.output_tokens` (OpenAI-compat streams require "
+                "`stream_options.include_usage=true`).",
+                extra={
+                    "model": response_data.get("model", "unknown"),
+                    "usd_cost": usd_cost,
+                    "usage_keys": sorted(usage_data.keys())
+                    if isinstance(usage_data, dict)
+                    else None,
+                },
+            )
+        try:
+            cost_details = usage_data.get("cost_details", {})
+            if not isinstance(cost_details, dict):
+                cost_details = {}
+            input_usd = _coerce_usd(
+                cost_details.get("input_cost")
+                or cost_details.get("upstream_inference_prompt_cost")
+            )
+            output_usd = _coerce_usd(
+                cost_details.get("output_cost")
+                or cost_details.get("upstream_inference_completions_cost")
+            )
+            return _calculate_from_usd_cost(
+                usd_cost,
+                input_usd,
+                output_usd,
+                input_tokens,
+                cache_read_tokens,
+                cache_creation_tokens,
+                output_tokens,
+                response_data,
             )
         except Exception as e:
             logger.warning(
@@ -114,69 +187,35 @@ async def calculate_cost(  # todo: can be sync
                     "model": response_data.get("model", "unknown"),
                 },
             )
-            # Fall through to token-based calculation
 
-    MSATS_PER_1K_INPUT_TOKENS: float = (
-        float(settings.fixed_per_1k_input_tokens) * 1000.0
-    )
-    MSATS_PER_1K_OUTPUT_TOKENS: float = (
-        float(settings.fixed_per_1k_output_tokens) * 1000.0
-    )
+    # Fall back to token-based pricing
+    try:
+        pricing_rates = _get_pricing_rates(response_data)
+    except ValueError as e:
+        return CostDataError(message=str(e), code="pricing_error")
 
-    if not settings.fixed_pricing:
-        response_model = response_data.get("model", "")
-        logger.debug(
-            "Using model-based pricing",
-            extra={"model": response_model},
-        )
+    if pricing_rates is None:
+        input_rate = float(settings.fixed_per_1k_input_tokens) * 1000.0
+        output_rate = float(settings.fixed_per_1k_output_tokens) * 1000.0
+        cache_read_rate = input_rate
+        cache_creation_rate = input_rate
+    else:
+        input_rate, output_rate, cache_read_rate, cache_creation_rate = pricing_rates
 
-        from ..proxy import get_model_instance
-
-        model_obj = get_model_instance(response_model)
-
-        if not model_obj:
-            logger.error(
-                "Invalid model in response",
-                extra={"response_model": response_model},
-            )
-            return CostDataError(
-                message=f"Invalid model in response: {response_model}",
-                code="model_not_found",
-            )
-
-        if not model_obj.sats_pricing:
-            logger.error(
-                "Model pricing not defined",
-                extra={"model": response_model, "model_id": response_model},
-            )
-            return CostDataError(
-                message="Model pricing not defined", code="pricing_not_found"
-            )
-
-        try:
-            mspp = float(model_obj.sats_pricing.prompt)
-            mspc = float(model_obj.sats_pricing.completion)
-        except Exception:
-            return CostDataError(message="Invalid pricing data", code="pricing_invalid")
-
-        MSATS_PER_1K_INPUT_TOKENS = mspp * 1_000_000.0
-        MSATS_PER_1K_OUTPUT_TOKENS = mspc * 1_000_000.0
-
-        logger.info(
-            "Applied model-specific pricing",
-            extra={
-                "model": response_model,
-                "input_price_msats_per_1k": MSATS_PER_1K_INPUT_TOKENS,
-                "output_price_msats_per_1k": MSATS_PER_1K_OUTPUT_TOKENS,
-            },
-        )
-
-    if not (MSATS_PER_1K_OUTPUT_TOKENS and MSATS_PER_1K_INPUT_TOKENS):
+    if not (input_rate and output_rate):
         logger.warning(
-            "No token pricing configured, using base cost",
+            "No token pricing configured — billing at flat MaxCostData. "
+            "Token counts %s in the upstream response but cannot be "
+            "priced; the request will appear in dashboards with the "
+            "raw counts and a fixed max-cost charge.",
+            "are present"
+            if (input_tokens > 0 or output_tokens > 0)
+            else "are zero",
             extra={
                 "base_cost_msats": max_cost,
                 "model": response_data.get("model", "unknown"),
+                "input_tokens": input_tokens,
+                "output_tokens": output_tokens,
             },
         )
         return MaxCostData(
@@ -184,35 +223,282 @@ async def calculate_cost(  # todo: can be sync
             input_msats=0,
             output_msats=0,
             total_msats=max_cost,
+            input_tokens=input_tokens,
+            output_tokens=output_tokens,
+            cache_read_input_tokens=cache_read_tokens,
+            cache_creation_input_tokens=cache_creation_tokens,
+            cache_read_msats=0,
+            cache_creation_msats=0,
         )
 
-    input_tokens = usage_data.get("prompt_tokens", 0)
-    output_tokens = usage_data.get("completion_tokens", 0)
-
-    # added for response api
-    input_tokens = (
-        input_tokens if input_tokens != 0 else usage_data.get("input_tokens", 0)
-    )
-    output_tokens = (
-        output_tokens if output_tokens != 0 else usage_data.get("output_tokens", 0)
+    return _calculate_from_tokens(
+        input_tokens,
+        output_tokens,
+        cache_read_tokens,
+        cache_creation_tokens,
+        input_rate,
+        output_rate,
+        cache_read_rate,
+        cache_creation_rate,
+        response_data,
     )
 
-    # added for response api
-    input_tokens = (
-        input_tokens
-        if input_tokens != 0
-        else response_data.get("usage", {}).get("input_tokens", 0)
+
+# ============================================================================
+# Helper Functions (ordered by call sequence in calculate_cost)
+# ============================================================================
+
+
+def _coerce_usd(value: object) -> float:
+    """Coerce a value to USD float, handling various formats safely."""
+    if value is None or isinstance(value, bool):
+        return 0.0
+    if not isinstance(value, (int, float, str)):
+        return 0.0
+    try:
+        return max(0.0, float(value))
+    except (TypeError, ValueError):
+        return 0.0
+
+
+def _resolve_usd_cost(usage_data: dict, response_data: dict) -> float:
+    """Resolve USD cost with clear priority order.
+
+    Priority:
+
+    1. ``cost_details.total_cost``
+    2. ``cost_details.upstream_inference_cost`` (BYOK — see below)
+    3. ``total_cost`` → ``cost`` (in both usage and response)
+
+    **BYOK path (PPQ.AI):** when ``is_byok`` is true the ``usage.cost`` field
+    is only a small (~5 %) routing fee, not the inference cost.  The real cost
+    lives in ``cost_details.upstream_inference_cost`` and the provider's
+    balance is debited by ``upstream_inference_cost + byok_fee``.  Billing just
+    the fee under-charges by ~20×.
+    """
+    cost_details = usage_data.get("cost_details")
+    if isinstance(cost_details, dict):
+        cost = _coerce_usd(cost_details.get("total_cost"))
+        if cost > 0:
+            return cost
+
+        # PPQ.AI BYOK: upstream_inference_cost is the real inference cost;
+        # usage.cost is only a ~5 % BYOK routing fee.  Bill the sum — what PPQ
+        # actually deducts from the balance.  For non-BYOK providers (e.g.
+        # OpenRouter) usage.cost already equals upstream_inference_cost, so we
+        # fall through to the normal ``cost`` lookup below.
+        upstream_cost = _coerce_usd(
+            cost_details.get("upstream_inference_cost")
+        )
+        if upstream_cost > 0 and usage_data.get("is_byok"):
+            byok_fee = _coerce_usd(usage_data.get("cost"))
+            return upstream_cost + byok_fee
+
+    for source in [usage_data, response_data]:
+        if not isinstance(source, dict):
+            continue
+        for field in ("total_cost", "cost"):
+            cost = _coerce_usd(source.get(field))
+            if cost > 0:
+                return cost
+
+    return 0.0
+
+
+def _get_pricing_rates(
+    response_data: dict,
+) -> tuple[float, float, float, float] | None:
+    """Get configured rates, falling back to LiteLLM's model cost map.
+
+    Returns: (input_rate, output_rate, cache_read_rate, cache_write_rate).
+    ``None`` means configured fixed pricing should be used by the caller.
+    """
+    if settings.fixed_pricing and (
+        settings.fixed_per_1k_input_tokens
+        or settings.fixed_per_1k_output_tokens
+    ):
+        return None
+
+    from ..proxy import get_model_instance
+    from .models import litellm_cost_entry
+
+    response_model = response_data.get("model", "")
+    model_obj = get_model_instance(response_model)
+
+    if model_obj and model_obj.sats_pricing:
+        try:
+            mspp = float(model_obj.sats_pricing.prompt)
+            mspc = float(model_obj.sats_pricing.completion)
+            mscr = float(model_obj.sats_pricing.input_cache_read or 0)
+            mscw = float(model_obj.sats_pricing.input_cache_write or 0)
+
+            mspp_1k = mspp * 1_000_000.0
+            mspc_1k = mspc * 1_000_000.0
+            mscr_1k = mscr * 1_000_000.0 if mscr > 0 else mspp_1k
+            mscw_1k = mscw * 1_000_000.0 if mscw > 0 else mspp_1k
+            source = "configured"
+        except Exception as e:
+            logger.error("Invalid pricing data", extra={"error": str(e)})
+            raise ValueError("Invalid pricing data") from e
+    else:
+        pricing_model = (
+            model_obj.forwarded_model_id if model_obj else None
+        ) or response_model
+        pricing = litellm_cost_entry(pricing_model)
+        if pricing is None:
+            logger.error(
+                "Model pricing not found in configured models or LiteLLM",
+                extra={
+                    "response_model": response_model,
+                    "pricing_model": pricing_model,
+                },
+            )
+            raise ValueError(f"Pricing not found for model: {response_model}")
+
+        input_usd = _coerce_usd(pricing.get("input_cost_per_token"))
+        output_usd = _coerce_usd(pricing.get("output_cost_per_token"))
+        if input_usd <= 0 or output_usd <= 0:
+            raise ValueError(f"Incomplete LiteLLM pricing for model: {pricing_model}")
+
+        provider_fee = _resolve_provider_fee(response_model)
+        usd_per_sat = sats_usd_price()
+        mspp_1k = input_usd * provider_fee * 1_000_000.0 / usd_per_sat
+        mspc_1k = output_usd * provider_fee * 1_000_000.0 / usd_per_sat
+        cache_read_usd = _coerce_usd(
+            pricing.get("cache_read_input_token_cost")
+        )
+        cache_write_usd = _coerce_usd(
+            pricing.get("cache_creation_input_token_cost")
+        )
+        mscr_1k = (
+            cache_read_usd * provider_fee * 1_000_000.0 / usd_per_sat
+            if cache_read_usd > 0
+            else mspp_1k
+        )
+        mscw_1k = (
+            cache_write_usd * provider_fee * 1_000_000.0 / usd_per_sat
+            if cache_write_usd > 0
+            else mspp_1k
+        )
+        source = "litellm"
+
+    logger.info(
+        "Applied model-specific pricing",
+        extra={
+            "model": response_model,
+            "pricing_source": source,
+            "input_price_msats_per_1k": mspp_1k,
+            "output_price_msats_per_1k": mspc_1k,
+            "cache_read_price_msats_per_1k": mscr_1k,
+            "cache_write_price_msats_per_1k": mscw_1k,
+        },
     )
-    output_tokens = (
-        output_tokens
-        if output_tokens != 0
-        else response_data.get("usage", {}).get("output_tokens", 0)
+    return mspp_1k, mspc_1k, mscr_1k, mscw_1k
+
+
+def _resolve_provider_fee(model_id: str) -> float:
+    """Resolve the provider fee multiplier for the given model id.
+
+    Falls back to 1.0 (no markup) when the provider cannot be resolved so
+    the USD cost path never silently double-applies or omits the fee.
+    """
+    from ..proxy import get_provider_for_model
+
+    if not model_id:
+        return 1.0
+    providers = get_provider_for_model(model_id)
+    if not providers:
+        return 1.0
+    return float(providers[0].provider_fee)
+
+
+def _calculate_from_usd_cost(
+    usd_cost: float,
+    input_usd: float,
+    output_usd: float,
+    input_tokens: int,
+    cache_read_tokens: int,
+    cache_creation_tokens: int,
+    output_tokens: int,
+    response_data: dict,
+) -> CostData:
+    """Calculate cost from USD figures, deriving input/output split from tokens."""
+    provider_fee = _resolve_provider_fee(response_data.get("model", ""))
+    usd_cost = usd_cost * provider_fee
+    input_usd = input_usd * provider_fee
+    output_usd = output_usd * provider_fee
+    sats_per_usd = 1.0 / sats_usd_price()
+    cost_in_sats = usd_cost * sats_per_usd
+    cost_in_msats = math.ceil(cost_in_sats * 1000)
+
+    if input_usd > 0 or output_usd > 0:
+        # The total is the authoritative billed amount. Allocating that integer
+        # total proportionally avoids losing sub-millisatoshi remainders when
+        # input and output components are each truncated independently.
+        component_usd = input_usd + output_usd
+        input_msats = math.floor(cost_in_msats * input_usd / component_usd)
+        output_msats = cost_in_msats - input_msats
+    else:
+        effective_input_tokens = (
+            input_tokens + cache_read_tokens + cache_creation_tokens
+        )
+        total_tokens = effective_input_tokens + output_tokens
+        input_msats = (
+            int(cost_in_msats * effective_input_tokens / total_tokens)
+            if total_tokens > 0
+            else 0
+        )
+        output_msats = cost_in_msats - input_msats
+
+    logger.info(
+        "Using cost from usage data/details",
+        extra={
+            "usd_cost": usd_cost,
+            "cost_in_sats": cost_in_sats,
+            "cost_in_msats": cost_in_msats,
+            "model": response_data.get("model", "unknown"),
+        },
     )
 
-    input_msats = round(input_tokens / 1000 * MSATS_PER_1K_INPUT_TOKENS, 3)
+    return CostData(
+        base_msats=0,
+        input_msats=input_msats,
+        output_msats=output_msats,
+        total_msats=cost_in_msats,
+        total_usd=usd_cost,
+        input_tokens=input_tokens,
+        output_tokens=output_tokens,
+        cache_read_input_tokens=cache_read_tokens,
+        cache_creation_input_tokens=cache_creation_tokens,
+        cache_read_msats=0,
+        cache_creation_msats=0,
+    )
 
-    output_msats = round(output_tokens / 1000 * MSATS_PER_1K_OUTPUT_TOKENS, 3)
-    token_based_cost = math.ceil(input_msats + output_msats)
+
+def _calculate_from_tokens(
+    input_tokens: int,
+    output_tokens: int,
+    cache_read_tokens: int,
+    cache_creation_tokens: int,
+    input_rate: float,
+    output_rate: float,
+    cache_read_rate: float,
+    cache_creation_rate: float,
+    response_data: dict,
+) -> CostData:
+    """Calculate cost from token counts using pricing rates."""
+    calc_input_msats = round(input_tokens / 1000 * input_rate, 3)
+    calc_output_msats = round(output_tokens / 1000 * output_rate, 3)
+    calc_cache_read_msats = round(cache_read_tokens / 1000 * cache_read_rate, 3)
+    calc_cache_write_msats = round(
+        cache_creation_tokens / 1000 * cache_creation_rate, 3
+    )
+    token_based_cost = math.ceil(
+        calc_input_msats
+        + calc_output_msats
+        + calc_cache_read_msats
+        + calc_cache_write_msats
+    )
     total_usd = (token_based_cost / 1000.0) * sats_usd_price()
 
     logger.info(
@@ -220,18 +506,39 @@ async def calculate_cost(  # todo: can be sync
         extra={
             "input_tokens": input_tokens,
             "output_tokens": output_tokens,
-            "input_cost_msats": input_msats,
-            "output_cost_msats": output_msats,
+            "cache_read_input_tokens": cache_read_tokens,
+            "cache_creation_input_tokens": cache_creation_tokens,
+            "input_cost_msats": calc_input_msats,
+            "output_cost_msats": calc_output_msats,
+            "cache_read_cost_msats": calc_cache_read_msats,
+            "cache_creation_cost_msats": calc_cache_write_msats,
             "total_cost_msats": token_based_cost,
             "total_usd": total_usd,
             "model": response_data.get("model", "unknown"),
         },
     )
 
+    # Fold the cache-read/write cost into the visible ``input_msats`` so a
+    # dashboard that renders I / O / T sees ``input + output == total``
+    # exactly. This mirrors ``_fold_cache_into_input_tokens`` (which rolls the
+    # cache token counts into the visible prompt total). The standalone
+    # ``cache_read_msats`` / ``cache_creation_msats`` fields stay populated for
+    # clients that want the breakdown; nothing sums the components to derive
+    # ``total_msats`` (it is computed independently above), so this is
+    # display-only and does not change what is billed.
+    visible_output_msats = int(calc_output_msats)
+    visible_input_msats = token_based_cost - visible_output_msats
+
     return CostData(
         base_msats=0,
-        input_msats=int(input_msats),
-        output_msats=int(output_msats),
+        input_msats=visible_input_msats,
+        output_msats=visible_output_msats,
         total_msats=token_based_cost,
         total_usd=total_usd,
+        input_tokens=input_tokens,
+        output_tokens=output_tokens,
+        cache_read_input_tokens=cache_read_tokens,
+        cache_creation_input_tokens=cache_creation_tokens,
+        cache_read_msats=int(calc_cache_read_msats),
+        cache_creation_msats=int(calc_cache_write_msats),
     )

routstr/payment/helpers.py

@@ -11,6 +11,8 @@ from PIL import Image
 from sqlmodel.ext.asyncio.session import AsyncSession
 
 from ..core import get_logger
+from ..core.exceptions import UpstreamError
+from ..core.redaction import redact_org_ids
 from ..core.settings import settings
 from ..wallet import deserialize_token_from_string
 
@@ -29,15 +31,13 @@ def check_token_balance(headers: dict, body: dict, max_cost_for_model: int) -> N
             },
         )
     elif auth := headers.get("authorization", None):
-        cashu_token = auth.split(" ")[1] if len(auth.split(" ")) > 1 else ""
         logger.debug(
-            "Using Authorization header token",
+            "Skipping preflight token balance check for Authorization header",
             extra={
-                "token_preview": cashu_token[:20] + "..."
-                if len(cashu_token) > 20
-                else cashu_token
+                "auth_preview": auth[:20] + "..." if len(auth) > 20 else auth,
             },
         )
+        return
     else:
         logger.error("No authentication token provided")
         raise HTTPException(status_code=401, detail="Unauthorized")
@@ -75,7 +75,7 @@ def check_token_balance(headers: dict, body: dict, max_cost_for_model: int) -> N
 
     if max_cost_for_model > amount_msat:
         raise HTTPException(
-            status_code=413,
+            status_code=402,
             detail={
                 "reason": "Insufficient balance",
                 "amount_required_msat": max_cost_for_model,
@@ -420,16 +420,27 @@ def create_error_response(
     status_code: int,
     request: Request,
     token: str | None = None,
+    code: str | int | None = None,
+    details: dict[str, object] | None = None,
 ) -> Response:
-    """Create a standardized error response."""
+    """Create a standardized error response.
+
+    ``code`` is a stable, machine-readable classification (e.g.
+    ``UPSTREAM_RATE_LIMIT``); when omitted it defaults to the HTTP status code
+    for backwards compatibility. ``details`` carries optional structured,
+    redaction-safe context.
+    """
+    error_obj: dict[str, object] = {
+        "message": redact_org_ids(message),
+        "type": error_type,
+        "code": code if code is not None else status_code,
+    }
+    if details is not None:
+        error_obj["details"] = details
     return Response(
         content=json.dumps(
             {
-                "error": {
-                    "message": message,
-                    "type": error_type,
-                    "code": status_code,
-                },
+                "error": error_obj,
                 "request_id": getattr(request.state, "request_id", "unknown"),
             }
         ),
@@ -437,3 +448,20 @@ def create_error_response(
         media_type="application/json",
         headers={"X-Cashu": token} if token else {},
     )
+
+
+def create_upstream_error_response(
+    error: UpstreamError,
+    request: Request,
+    fallback_status: int = 502,
+) -> Response:
+    """Build an error response from an :class:`UpstreamError`, preserving its
+    structured ``code``, ``details``, and original ``status_code``."""
+    return create_error_response(
+        "upstream_error",
+        str(error),
+        error.status_code or fallback_status,
+        request=request,
+        code=getattr(error, "code", None),
+        details=getattr(error, "details", None),
+    )

routstr/payment/lnurl.py

@@ -1,11 +1,17 @@
 from __future__ import annotations
 
+import asyncio
 import math
 from typing import TypedDict
 
 import httpx
 from cashu.wallet.wallet import Proof, Wallet
 
+# The Cashu library issues POST /v1/melt/bolt11 with timeout=None, so a hung or
+# very slow mint can block a melt (and any caller, e.g. the payout loop)
+# indefinitely. Bound it here so callers fail instead of hanging forever.
+MELT_TIMEOUT_SECONDS = 60
+
 try:
     from bech32 import bech32_decode, convertbits  # type: ignore
 except ModuleNotFoundError:  # pragma: no cover – allow runtime miss
@@ -220,10 +226,18 @@ async def raw_send_to_lnurl(
     if amount:
         proofs, _ = await wallet.select_to_send(proofs, amount, set_reserved=True)
 
-    _ = await wallet.melt(
-        proofs=proofs,
-        invoice=bolt11_invoice,
-        fee_reserve_sat=melt_quote_resp.fee_reserve,
-        quote_id=melt_quote_resp.quote,
-    )
+    try:
+        _ = await asyncio.wait_for(
+            wallet.melt(
+                proofs=proofs,
+                invoice=bolt11_invoice,
+                fee_reserve_sat=melt_quote_resp.fee_reserve,
+                quote_id=melt_quote_resp.quote,
+            ),
+            timeout=MELT_TIMEOUT_SECONDS,
+        )
+    except asyncio.TimeoutError as e:
+        raise LNURLError(
+            f"Melt timed out after {MELT_TIMEOUT_SECONDS}s (mint unresponsive)"
+        ) from e
     return final_amount

routstr/payment/models.py

@@ -3,11 +3,12 @@ import json
 import random
 
 import httpx
-from fastapi import APIRouter, Depends
+from fastapi import APIRouter, Depends, HTTPException, Request
+from pydantic import BaseModel as V2BaseModel
 from pydantic.v1 import BaseModel
 from sqlmodel.ext.asyncio.session import AsyncSession
 
-from ..core.db import ModelRow, get_session
+from ..core.db import ModelRow, UpstreamProviderRow, get_session
 from ..core.logging import get_logger
 from ..core.settings import settings
 from .price import sats_usd_price
@@ -16,6 +17,24 @@ logger = get_logger(__name__)
 
 models_router = APIRouter()
 
+_MODEL_TEST_ENDPOINT_PATHS = {
+    "chat-completions": "chat/completions",
+    "completions": "completions",
+    "embeddings": "embeddings",
+    "responses": "responses",
+}
+
+# Cap the caller-supplied test payload to avoid forwarding oversized bodies
+# upstream on the operator's credentials.
+_MODEL_TEST_MAX_REQUEST_BYTES = 64 * 1024
+
+
+async def _require_admin_api(request: Request) -> None:
+    """Require admin auth without creating an import-time cycle with core.admin."""
+    from ..core.admin import require_admin_api
+
+    await require_admin_api(request)
+
 
 class Architecture(BaseModel):
     modality: str
@@ -60,11 +79,70 @@ class Model(BaseModel):
     upstream_provider_id: int | str | None = None
     canonical_slug: str | None = None
     alias_ids: list[str] | None = None
+    forwarded_model_id: str | None = None
 
     def __hash__(self) -> int:
         return hash(self.id)
 
 
+def litellm_cost_entry(model_id: str) -> dict | None:
+    """Look up ``model_id`` in litellm's bundled cost map.
+
+    litellm ships per-model USD rates keyed by the exact OpenRouter id
+    (``deepseek/deepseek-chat``) or the bare model name (``gpt-4o``,
+    ``claude-sonnet-4-5``), so both spellings are tried. Keys are lowercase, so
+    a mixed-case upstream id (``deepseek-ai/DeepSeek-V4-Flash``) is retried via
+    a case-insensitive scan. Returns the matched cost dict, or ``None``.
+    """
+    import litellm
+
+    candidates = (model_id, model_id.split("/", 1)[-1])
+    for key in candidates:
+        info = litellm.model_cost.get(key)
+        if isinstance(info, dict):
+            return info
+
+    lowered = {c.lower() for c in candidates}
+    for key, info in litellm.model_cost.items():
+        if isinstance(key, str) and key.lower() in lowered and isinstance(info, dict):
+            return info
+    return None
+
+
+def backfill_cache_pricing(model_id: str, pricing: Pricing) -> Pricing:
+    """Fill missing cache rates from litellm's bundled cost map.
+
+    The OpenRouter model feed omits ``input_cache_read``/``input_cache_write``
+    for many models (most DeepSeek entries, openai/gpt-4o, ...). Without a
+    cache rate, billing falls back to the full input rate, which overcharges
+    cache reads (DeepSeek hits are 10x cheaper) and undercharges Anthropic
+    cache writes (1.25x). The lookup (see ``litellm_cost_entry``) tries both
+    id spellings and a case-insensitive fallback.
+
+    Rates already present (e.g. provided by OpenRouter) are authoritative and
+    never overwritten. Unknown models are returned unchanged.
+    """
+    needs_read = (pricing.input_cache_read or 0.0) <= 0.0
+    needs_write = (pricing.input_cache_write or 0.0) <= 0.0
+    if not (needs_read or needs_write):
+        return pricing
+
+    info = litellm_cost_entry(model_id)
+    if info is None:
+        return pricing
+
+    updated = Pricing.parse_obj(pricing.dict())
+    if needs_read:
+        read_rate = info.get("cache_read_input_token_cost")
+        if isinstance(read_rate, (int, float)) and read_rate > 0:
+            updated.input_cache_read = float(read_rate)
+    if needs_write:
+        write_rate = info.get("cache_creation_input_token_cost")
+        if isinstance(write_rate, (int, float)) and write_rate > 0:
+            updated.input_cache_write = float(write_rate)
+    return updated
+
+
 def _has_valid_pricing(model: dict) -> bool:
     """Check if model has valid pricing (not free, no negative values)."""
     pricing = model.get("pricing", {})
@@ -153,13 +231,29 @@ def _row_to_model(
     )
     top_provider_dict = json.loads(row.top_provider) if row.top_provider else None
 
-    if apply_provider_fee and isinstance(pricing, dict):
-        pricing = {k: float(v) * provider_fee for k, v in pricing.items()}
-
     if isinstance(pricing, dict) and float(pricing.get("request", 0.0)) <= 0.0:
         pricing["request"] = max(pricing.get("request", 0.0), 0.0)
 
     parsed_pricing = Pricing.parse_obj(pricing)
+
+    # Fill missing cache-read/write rates from litellm's cost map BEFORE applying
+    # the provider fee, so they carry the same markup as every other component.
+    # DB-stored override pricing (e.g. generic providers) omits cache rates;
+    # without this, ``_row_to_model`` bills cache reads at the full input rate —
+    # the ``_apply_provider_fee_to_model`` path backfills, but the override path
+    # used for admin-configured providers did not.
+    #
+    # Key on ``forwarded_model_id`` (the actual upstream model name litellm
+    # prices) when set: an alias row (id="local-alias",
+    # forwarded_model_id="deepseek-v4-flash") would otherwise look up the alias
+    # and miss the cache rate.
+    pricing_model_id = getattr(row, "forwarded_model_id", None) or row.id
+    parsed_pricing = backfill_cache_pricing(pricing_model_id, parsed_pricing)
+
+    if apply_provider_fee:
+        parsed_pricing = Pricing.parse_obj(
+            {k: float(v) * provider_fee for k, v in parsed_pricing.dict().items()}
+        )
     model = Model(
         id=row.id,
         name=row.name,
@@ -177,6 +271,7 @@ def _row_to_model(
         upstream_provider_id=row.upstream_provider_id,
         canonical_slug=getattr(row, "canonical_slug", None),
         alias_ids=json.loads(row.alias_ids) if row.alias_ids else None,
+        forwarded_model_id=getattr(row, "forwarded_model_id", None) or row.id,
     )
 
     if apply_provider_fee:
@@ -329,6 +424,7 @@ def _update_model_sats_pricing(model: Model, sats_to_usd: float) -> Model:
             upstream_provider_id=model.upstream_provider_id,
             canonical_slug=model.canonical_slug,
             alias_ids=model.alias_ids,
+            forwarded_model_id=model.forwarded_model_id,
         )
     except Exception as e:
         logger.error(
@@ -347,6 +443,9 @@ async def _update_sats_pricing_once() -> None:
     from ..proxy import get_upstreams, refresh_model_maps
 
     upstreams = get_upstreams()
+    if not upstreams:
+        return
+
     sats_to_usd = sats_usd_price()
 
     updated_count = 0
@@ -356,11 +455,14 @@ async def _update_sats_pricing_once() -> None:
             for m in upstream.get_cached_models()
         ]
         upstream._models_cache = updated_models
-        upstream._models_by_id = {m.id: m for m in updated_models}
+        upstream._models_by_id = {m.forwarded_model_id or m.id: m for m in updated_models}
         updated_count += len(updated_models)
 
     if updated_count > 0:
-        logger.info("Updated sats pricing", extra={"models_updated": updated_count})
+        logger.info(
+            f"Updated sats pricing for {updated_count} models",
+            extra={"models_updated": updated_count},
+        )
         await refresh_model_maps()
 
 
@@ -402,11 +504,110 @@ async def update_sats_pricing() -> None:
             logger.error(f"Error updating sats pricing: {e}")
 
 
+class ModelTestRequest(V2BaseModel):
+    model_id: str
+    endpoint_type: str
+    request_data: dict
+
+
+@models_router.post(
+    "/api/models/test", dependencies=[Depends(_require_admin_api)]
+)
+async def test_model(
+    payload: ModelTestRequest,
+    session: AsyncSession = Depends(get_session),
+) -> dict:
+    """Test a model by sending a request through its configured upstream provider."""
+    from sqlmodel import select
+
+    result = await session.execute(
+        select(ModelRow).where(ModelRow.id == payload.model_id)
+    )
+    model_row = result.scalars().first()
+
+    if not model_row:
+        return {
+            "success": False,
+            "error": f"Model '{payload.model_id}' not found in database",
+            "status_code": 404,
+        }
+
+    provider = await session.get(UpstreamProviderRow, model_row.upstream_provider_id)
+    if not provider:
+        return {
+            "success": False,
+            "error": "Upstream provider not found",
+            "status_code": 404,
+        }
+
+    endpoint_path = _MODEL_TEST_ENDPOINT_PATHS.get(payload.endpoint_type)
+    if endpoint_path is None:
+        raise HTTPException(status_code=400, detail="Unsupported endpoint_type")
+
+    actual_model_id = model_row.forwarded_model_id or model_row.id
+    request_data = dict(payload.request_data)
+    request_data["model"] = actual_model_id
+
+    try:
+        request_size = len(json.dumps(request_data).encode("utf-8"))
+    except (TypeError, ValueError):
+        raise HTTPException(status_code=400, detail="Invalid request_data")
+    if request_size > _MODEL_TEST_MAX_REQUEST_BYTES:
+        raise HTTPException(status_code=413, detail="request_data too large")
+
+    base_url = provider.base_url.rstrip("/")
+    url = f"{base_url}/{endpoint_path}"
+
+    logger.info(
+        "admin model test",
+        extra={
+            "model_id": payload.model_id,
+            "forwarded_model_id": actual_model_id,
+            "endpoint_type": payload.endpoint_type,
+            "upstream_provider_id": model_row.upstream_provider_id,
+            "request_bytes": request_size,
+        },
+    )
+
+    headers = {
+        "Content-Type": "application/json",
+        "Authorization": f"Bearer {provider.api_key}",
+    }
+
+    try:
+        async with httpx.AsyncClient(timeout=30.0) as client:
+            response = await client.post(url, json=request_data, headers=headers)
+            try:
+                response_data = response.json()
+            except Exception:
+                response_data = {"raw": response.text}
+
+            return {
+                "success": response.status_code < 400,
+                "data": response_data,
+                "status_code": response.status_code,
+            }
+    except Exception as e:
+        return {
+            "success": False,
+            "error": str(e),
+            "status_code": 500,
+        }
+
+
 @models_router.get("/v1/models")
-@models_router.get("/models", include_in_schema=False)
+@models_router.get("/v1/models/", include_in_schema=False)
+@models_router.get("/models")
+@models_router.get("/models/", include_in_schema=False)
 async def models(session: AsyncSession = Depends(get_session)) -> dict:
     """Get all available models from all providers with database overrides applied."""
     from ..proxy import get_unique_models
 
     items = get_unique_models()
-    return {"data": items}
+    data = []
+    for model in items:
+        m = model.dict()
+        if model.forwarded_model_id:
+            m["id"] = model.forwarded_model_id
+        data.append(m)
+    return {"data": data}

routstr/payment/usage.py

@@ -0,0 +1,131 @@
+"""Vendor-agnostic normalization of upstream usage objects.
+
+Upstream providers report token usage in vendor dialects that differ in field
+names and in whether cached tokens are included in the input count:
+
+* OpenAI / Azure / xAI / Groq / Moonshot / Qwen / Gemini-compat: cache reads in
+  ``prompt_tokens_details.cached_tokens``, included in ``prompt_tokens``.
+* OpenRouter: same as OpenAI plus cache *writes* in
+  ``prompt_tokens_details.cache_write_tokens``, also included in
+  ``prompt_tokens``.
+* litellm-normalized: same nesting, but names the write field
+  ``prompt_tokens_details.cache_creation_tokens`` (and additionally mirrors the
+  Anthropic top-level fields), with ``prompt_tokens`` as the grand total.
+* Anthropic native: ``cache_read_input_tokens`` / ``cache_creation_input_tokens``
+  top-level, additive to (not included in) ``input_tokens``.
+* DeepSeek: ``prompt_cache_hit_tokens`` / ``prompt_cache_miss_tokens``, with
+  ``prompt_tokens = hit + miss``.
+
+What decides whether cached tokens must be subtracted out of the input count is
+**which prompt field the vendor uses**, not which cache field appears:
+
+* ``prompt_tokens`` present -> cached + cache-write tokens are *included* in it
+  (OpenAI family, DeepSeek, OpenRouter, litellm); subtract both so
+  ``input_tokens`` holds only the regular-rate portion.
+* only ``input_tokens`` (Anthropic native) -> cached tokens are *additive*;
+  leave ``input_tokens`` untouched.
+
+``normalize_usage`` maps all of them onto one canonical ``NormalizedUsage``
+shape so billing code needs no vendor knowledge. The known dialects' field
+names do not collide, so a single union parser is safe; a vendor whose fields
+would genuinely conflict needs a dedicated branch here.
+"""
+
+from pydantic.v1 import BaseModel
+
+
+class NormalizedUsage(BaseModel):
+    """Canonical token usage: input_tokens never includes cached tokens."""
+
+    input_tokens: int = 0
+    output_tokens: int = 0
+    cache_read_tokens: int = 0
+    cache_write_tokens: int = 0
+
+
+def parse_token_count(value: object) -> int:
+    """Parse a token count from various formats (int, float, str, bool)."""
+    if isinstance(value, bool):
+        return 0
+    if isinstance(value, int):
+        return max(0, value)
+    if isinstance(value, float):
+        return max(0, int(value))
+    if isinstance(value, str):
+        try:
+            return max(0, int(float(value)))
+        except ValueError:
+            return 0
+    return 0
+
+
+def _first_token_count(usage_data: dict, *fields: str) -> int:
+    """Return the first positive token count among the given fields."""
+    for field in fields:
+        value = parse_token_count(usage_data.get(field, 0))
+        if value > 0:
+            return value
+    return 0
+
+
+def _extract_cache_tokens(usage_data: dict) -> tuple[int, int]:
+    """Pull (cache_read, cache_write) across all known dialects.
+
+    Precedence (highest first), independent for reads and writes:
+
+    * Anthropic top-level: ``cache_read_input_tokens`` /
+      ``cache_creation_input_tokens``.
+    * Nested ``prompt_tokens_details``: ``cached_tokens`` for reads;
+      ``cache_creation_tokens`` (litellm) or ``cache_write_tokens``
+      (OpenRouter) for writes.
+    * DeepSeek: ``prompt_cache_hit_tokens`` for reads (no write concept).
+    """
+    cache_read = parse_token_count(usage_data.get("cache_read_input_tokens", 0))
+    cache_write = parse_token_count(usage_data.get("cache_creation_input_tokens", 0))
+
+    prompt_details = usage_data.get("prompt_tokens_details")
+    if isinstance(prompt_details, dict):
+        if not cache_read:
+            cache_read = parse_token_count(prompt_details.get("cached_tokens", 0))
+        if not cache_write:
+            cache_write = _first_token_count(
+                prompt_details, "cache_creation_tokens", "cache_write_tokens"
+            )
+
+    if not cache_read:
+        # DeepSeek: prompt_tokens = prompt_cache_hit_tokens + prompt_cache_miss_tokens
+        cache_read = parse_token_count(usage_data.get("prompt_cache_hit_tokens", 0))
+
+    return cache_read, cache_write
+
+
+def normalize_usage(usage_data: object) -> NormalizedUsage | None:
+    """Map a vendor usage dict onto the canonical shape, or None if absent.
+
+    Cached reads and writes are subtracted from the input count exactly once,
+    only for dialects that report a ``prompt_tokens`` grand total that already
+    includes them (OpenAI family, DeepSeek, OpenRouter, litellm). Anthropic
+    native reports them additively under ``input_tokens`` and is left untouched.
+    """
+    if not isinstance(usage_data, dict):
+        return None
+
+    output_tokens = _first_token_count(
+        usage_data, "completion_tokens", "output_tokens"
+    )
+    cache_read, cache_write = _extract_cache_tokens(usage_data)
+
+    # ``prompt_tokens`` is the inclusive grand total; ``input_tokens`` (Anthropic
+    # native) excludes cached tokens. The field chosen decides whether to subtract.
+    if "prompt_tokens" in usage_data:
+        input_tokens = parse_token_count(usage_data.get("prompt_tokens", 0))
+        input_tokens = max(0, input_tokens - cache_read - cache_write)
+    else:
+        input_tokens = parse_token_count(usage_data.get("input_tokens", 0))
+
+    return NormalizedUsage(
+        input_tokens=input_tokens,
+        output_tokens=output_tokens,
+        cache_read_tokens=cache_read,
+        cache_write_tokens=cache_write,
+    )

routstr/proxy.py

@@ -1,3 +1,4 @@
+import asyncio
 import json
 from typing import Any
 
@@ -17,16 +18,20 @@ from .core.db import (
     get_session,
 )
 from .core.exceptions import UpstreamError
+from .core.not_found import build_not_found_response
 from .core.settings import settings
 from .payment.helpers import (
     calculate_discounted_max_cost,
     check_token_balance,
     create_error_response,
+    create_upstream_error_response,
     get_max_cost_for_model,
 )
 from .payment.models import Model
 from .upstream import BaseUpstreamProvider
+from .upstream.ehbp import forward_ehbp_request, forward_ehbp_x_cashu_request
 from .upstream.helpers import init_upstreams
+from .upstream.request_correction import correct_request, extract_error_message
 
 logger = get_logger(__name__)
 proxy_router = APIRouter()
@@ -69,7 +74,25 @@ def get_upstreams() -> list[BaseUpstreamProvider]:
 
 def get_model_instance(model_id: str) -> Model | None:
     """Get Model instance by ID from global cache."""
-    return _model_instances.get(model_id.lower())
+    if not model_id:
+        return None
+
+    model_id_lower = model_id.lower()
+    # Try exact match first
+    if model := _model_instances.get(model_id_lower):
+        return model
+
+    # Try stripping common version suffixes (e.g., -20251222)
+    # This handles cases where upstream returns a specific version
+    # but we only track the base model name.
+    import re
+
+    base_model_id = re.sub(r"-\d{8}$", "", model_id_lower)
+    if base_model_id != model_id_lower:
+        if model := _model_instances.get(base_model_id):
+            return model
+
+    return None
 
 
 def get_provider_for_model(model_id: str) -> list[BaseUpstreamProvider] | None:
@@ -82,6 +105,34 @@ def get_unique_models() -> list[Model]:
     return list(_unique_models.values())
 
 
+def _is_tinfoil_attestation_path(path: str) -> bool:
+    """Return True for exact Tinfoil attestation routes, with optional slash."""
+    return path in {
+        "attestation",
+        "attestation/",
+        "tee/attestation",
+        "tee/attestation/",
+    }
+
+
+def _select_unauthenticated_get_upstreams(
+    path: str, upstreams: list[BaseUpstreamProvider]
+) -> list[BaseUpstreamProvider]:
+    """Select upstream candidates for unauthenticated GET bypass paths.
+
+    Tinfoil attestation endpoints are provider-specific. Trying every enabled
+    upstream can return an unrelated provider's 404 before Tinfoil is reached,
+    so route those paths only to Tinfoil providers.
+    """
+    if _is_tinfoil_attestation_path(path):
+        return [
+            upstream
+            for upstream in upstreams
+            if getattr(upstream, "provider_type", None) == "tinfoil"
+        ]
+    return upstreams
+
+
 async def refresh_model_maps() -> None:
     """Refresh global model and provider maps using the cost-based algorithm."""
     from sqlalchemy.orm import selectinload
@@ -96,22 +147,23 @@ async def refresh_model_maps() -> None:
         result = await session.exec(query)
         provider_rows = result.all()
 
-    overrides_by_id: dict[str, tuple[ModelRow, float]] = {}
-    disabled_model_ids: set[str] = set()
+    overrides_by_key: dict[tuple[str, int], tuple[ModelRow, float]] = {}
+    disabled_model_keys: set[tuple[str, int]] = set()
 
     for provider in provider_rows:
         if not provider.enabled:
             continue
         for model in provider.models:
+            model_key = (model.id.lower(), model.upstream_provider_id)
             if model.enabled:
-                overrides_by_id[model.id] = (model, provider.provider_fee)
+                overrides_by_key[model_key] = (model, provider.provider_fee)
             else:
-                disabled_model_ids.add(model.id)
+                disabled_model_keys.add(model_key)
 
     _model_instances, _provider_map, _unique_models = create_model_mappings(
         upstreams=_upstreams,
-        overrides_by_id=overrides_by_id,
-        disabled_model_ids=disabled_model_ids,
+        overrides_by_key=overrides_by_key,
+        disabled_model_keys=disabled_model_keys,
     )
 
 
@@ -132,27 +184,107 @@ async def refresh_model_maps_periodically() -> None:
             )
 
 
+_API_PATH_PREFIXES = (
+    "v1/",
+    "responses",
+    "chat/",
+    "completions",
+    "models",
+    "embeddings",
+    "audio/",
+    "images/",
+    "moderations",
+    "providers",
+    "tee/",
+    "attestation",
+)
+
+
 @proxy_router.api_route("/{path:path}", methods=["GET", "POST"], response_model=None)
 async def proxy(
     request: Request, path: str, session: AsyncSession = Depends(get_session)
 ) -> Response | StreamingResponse:
-    headers = dict(request.headers)
+    # GET requests must hit a known API prefix; otherwise return a 404 (HTML
+    # for browsers, JSON for API clients). POST requests are always forwarded
+    # so that OpenAI-style endpoints work with or without the `v1/` prefix
+    # (e.g. `/chat/completions` as well as `/v1/chat/completions`).
+    if request.method == "GET" and not path.startswith(_API_PATH_PREFIXES):
+        return build_not_found_response(request, path)
 
-    if "x-cashu" not in headers and "authorization" not in headers.keys():
-        return create_error_response(
-            "unauthorized", "Unauthorized", 401, request=request
-        )
+    headers = dict(request.headers)
 
     is_responses_api = path.startswith("v1/responses") or path.startswith("responses")
     request_body = await request.body()
-    request_body_dict = parse_request_body_json(request_body, path)
 
-    if is_responses_api:
-        model_id = extract_model_from_responses_request(request_body_dict)
+    # EHBP (Encrypted HTTP Body Protocol) requests carry an Ehbp-Encapsulated-Key
+    # header and a binary HPKE-sealed body. The proxy cannot parse the body to
+    # extract the model id, so the SDK sends it in X-Routstr-Model. Forward the
+    # raw encrypted body to the upstream's /private/ endpoint and stream the
+    # encrypted response back untouched — the SDK's SecureClient decrypts it.
+    is_ehbp = "ehbp-encapsulated-key" in headers
+    if is_ehbp:
+        request_body_dict = {}
+        model_id = headers.get("x-routstr-model", "")
+        if not model_id:
+            return create_error_response(
+                "invalid_request",
+                "EHBP request missing X-Routstr-Model header",
+                400,
+                request=request,
+            )
     else:
-        model_id = request_body_dict.get("model", "unknown")
+        request_body_dict = parse_request_body_json(request_body, path)
+        if is_responses_api:
+            model_id = extract_model_from_responses_request(request_body_dict)
+        else:
+            model_id = request_body_dict.get("model", "unknown")
+
+    # Exact Tinfoil attestation GET routes don't map to models — forward
+    # without model/cost/auth lookups. Do not prefix-match here: paths such as
+    # /attestationjunk must continue through normal authentication.
+    if request.method == "GET" and _is_tinfoil_attestation_path(path):
+        selected_upstreams = _select_unauthenticated_get_upstreams(path, _upstreams)
+        if not selected_upstreams:
+            return create_error_response(
+                "upstream_error",
+                "No upstream available for unauthenticated GET path",
+                502,
+                request=request,
+            )
+
+        last_error_response = None
+        for i, upstream in enumerate(selected_upstreams):
+            try:
+                headers = upstream.prepare_headers(dict(request.headers))
+                response = await upstream.forward_get_request(request, path, headers)
+                if (
+                    response.status_code in [502, 429]
+                    and i < len(selected_upstreams) - 1
+                ):
+                    logger.warning(
+                        "Upstream %s returned %s for unauthenticated GET %s, trying next",
+                        upstream.provider_type,
+                        response.status_code,
+                        path,
+                    )
+                    continue
+                return response
+            except UpstreamError as e:
+                logger.warning(
+                    "Upstream %s failed for unauthenticated GET %s: %s",
+                    upstream.provider_type,
+                    path,
+                    e,
+                )
+                if i == len(selected_upstreams) - 1:
+                    last_error_response = create_upstream_error_response(e, request)
+                continue
+        return last_error_response or create_error_response(
+            "upstream_error", "All upstreams failed", 502, request=request
+        )
 
     model_obj = get_model_instance(model_id)
+
     if not model_obj:
         return create_error_response(
             "invalid_model", f"Model '{model_id}' not found", 400, request=request
@@ -167,6 +299,16 @@ async def proxy(
             request=request,
         )
 
+    if is_ehbp:
+        upstreams = [upstream for upstream in upstreams if upstream.supports_ehbp]
+        if not upstreams:
+            return create_error_response(
+                "unsupported_request",
+                f"No EHBP-capable provider found for model '{model_id}'",
+                400,
+                request=request,
+            )
+
     # todo figure out cost calculation since fallback provider is usually not the same price
     # Use first provider for initial checks/cost calculation
     # primary_upstream = upstreams[0]
@@ -186,7 +328,23 @@ async def proxy(
         last_error = None
         for i, upstream in enumerate(upstreams):
             try:
-                if is_responses_api:
+                if is_ehbp:
+                    if not upstream.supports_ehbp:
+                        logger.warning(
+                            "Upstream %s does not support EHBP for model=%s",
+                            upstream.provider_type,
+                            model_id,
+                        )
+                        continue
+                    return await forward_ehbp_x_cashu_request(
+                        request=request,
+                        x_cashu_token=x_cashu,
+                        path=path,
+                        max_cost_for_model=max_cost_for_model,
+                        model_obj=model_obj,
+                        upstream=upstream,
+                    )
+                elif is_responses_api:
                     return await upstream.handle_x_cashu_responses(
                         request, x_cashu, path, max_cost_for_model, model_obj
                     )
@@ -196,22 +354,29 @@ async def proxy(
                     )
             except UpstreamError as e:
                 logger.warning(
-                    f"Upstream {upstream.provider_type} failed (x-cashu): {e}"
+                    "Upstream %s failed (x-cashu) for model=%s: %s",
+                    upstream.provider_type,
+                    model_id,
+                    e,
+                    extra={
+                        "provider": upstream.provider_type,
+                        "model": model_id,
+                        "status_code": e.status_code,
+                    },
                 )
                 if i == len(upstreams) - 1:
                     last_error = e
                 continue
 
+        if last_error is not None:
+            return create_upstream_error_response(last_error, request)
         return create_error_response(
-            "upstream_error",
-            str(last_error) if last_error else "All upstreams failed",
-            502,
-            request=request,
+            "upstream_error", "All upstreams failed", 502, request=request
         )
 
     elif auth := headers.get("authorization", None):
         key = await get_bearer_token_key(
-            headers, path, session, auth, max_cost_for_model
+            headers, path, session, auth, max_cost_for_model, model_id
         )
 
     else:
@@ -262,57 +427,117 @@ async def proxy(
             except UpstreamError as e:
                 logger.warning(f"Upstream {upstream.provider_type} failed (GET): {e}")
                 if i == len(upstreams) - 1:
-                    last_error_response = create_error_response(
-                        "upstream_error", str(e), 502, request=request
-                    )
+                    last_error_response = create_upstream_error_response(e, request)
                 continue
         return last_error_response or create_error_response(
             "upstream_error", "All upstreams failed", 502, request=request
         )
 
-    if request_body_dict:
+    if is_ehbp or request_body_dict:
         await pay_for_request(key, max_cost_for_model, session)
 
+    # Tracks request params already removed in response to upstream rejections,
+    # shared across providers so a stripped param stays stripped on failover and
+    # the reactive retry can never loop unboundedly.
+    already_stripped: set[str] = set()
+
     for i, upstream in enumerate(upstreams):
         headers = upstream.prepare_headers(dict(request.headers))
 
         try:
-            try:
-                if is_responses_api:
-                    response = await upstream.forward_responses_request(
-                        request,
-                        path,
-                        headers,
-                        request_body,
-                        key,
-                        max_cost_for_model,
-                        session,
-                        model_obj,
+            while True:
+                try:
+                    if is_ehbp:
+                        if not upstream.supports_ehbp:
+                            logger.warning(
+                                "Upstream %s does not support EHBP for model=%s",
+                                upstream.provider_type,
+                                model_id,
+                            )
+                            raise UpstreamError(
+                                f"Provider {upstream.provider_type} does not support EHBP",
+                                status_code=400,
+                            )
+                        response = await forward_ehbp_request(
+                            request=request,
+                            path=path,
+                            headers=headers,
+                            request_body=request_body,
+                            upstream=upstream,
+                            key=key,
+                            max_cost_for_model=max_cost_for_model,
+                            session=session,
+                            model_obj=model_obj,
+                        )
+                    elif is_responses_api:
+                        response = await upstream.forward_responses_request(
+                            request,
+                            path,
+                            headers,
+                            request_body,
+                            key,
+                            max_cost_for_model,
+                            session,
+                            model_obj,
+                        )
+                    else:
+                        response = await upstream.forward_request(
+                            request,
+                            path,
+                            headers,
+                            request_body,
+                            key,
+                            max_cost_for_model,
+                            session,
+                            model_obj,
+                        )
+                except UpstreamError:
+                    # Let the outer UpstreamError handler manage retry/revert
+                    raise
+                except Exception as e:
+                    # Unexpected error (not an upstream failure) — revert and propagate
+                    logger.error(
+                        "Unexpected error in upstream request, reverting payment",
+                        extra={
+                            "error": str(e),
+                            "error_type": type(e).__name__,
+                            "path": path,
+                            "key_hash": key.hashed_key[:8] + "...",
+                            "max_cost_for_model": max_cost_for_model,
+                        },
                     )
-                else:
-                    response = await upstream.forward_request(
-                        request,
-                        path,
-                        headers,
+                    await revert_pay_for_request(key, session, max_cost_for_model)
+                    raise
+
+                # Reactive recovery: some models reject one specific request
+                # param (e.g. newer Anthropic models deprecating `temperature`).
+                # When the upstream 400s naming such a param, strip it from the
+                # body and retry the SAME upstream. ``already_stripped`` bounds
+                # this to one retry per distinct param so it always terminates.
+                if response.status_code == 400 and not is_ehbp:
+                    correction = correct_request(
                         request_body,
-                        key,
-                        max_cost_for_model,
-                        session,
-                        model_obj,
+                        extract_error_message(response),
+                        already_stripped,
                     )
-            except Exception as e:
-                logger.error(
-                    "Upstream request failed, ensuring payment is reverted",
-                    extra={
-                        "error": str(e),
-                        "error_type": type(e).__name__,
-                        "path": path,
-                        "key_hash": key.hashed_key[:8] + "...",
-                        "max_cost_for_model": max_cost_for_model,
-                    },
-                )
-                await revert_pay_for_request(key, session, max_cost_for_model)
-                raise
+                    if correction is not None:
+                        request_body, bad_param = correction.body, correction.label
+                        already_stripped.add(bad_param)
+                        logger.warning(
+                            "Upstream %s rejected param '%s' for model=%s; "
+                            "stripping and retrying same upstream",
+                            upstream.provider_type,
+                            bad_param,
+                            model_id,
+                            extra={
+                                "provider": upstream.provider_type,
+                                "model": model_id,
+                                "stripped_param": bad_param,
+                                "path": path,
+                            },
+                        )
+                        continue
+                break
 
             if response.status_code != 200:
                 # Check if we should retry (502 Upstream Error or 429 Rate Limit)
@@ -337,10 +562,14 @@ async def proxy(
                     )
 
                     logger.warning(
-                        f"Upstream {upstream.provider_type} returned {response.status_code}, trying next provider",
+                        "Upstream %s returned %s for model=%s, trying next provider",
+                        upstream.provider_type,
+                        response.status_code,
+                        model_id,
                         extra={
                             "status_code": response.status_code,
-                            "upstream": upstream.provider_type,
+                            "provider": upstream.provider_type,
+                            "model": model_id,
                         },
                     )
                     continue
@@ -348,34 +577,59 @@ async def proxy(
                 # 4xx error (user error), or other non-retryable error, or last provider failed
                 await revert_pay_for_request(key, session, max_cost_for_model)
                 logger.warning(
-                    "Upstream request failed, revert payment",
+                    "Upstream request failed, revert payment "
+                    "(provider=%s model=%s status=%s path=%s)",
+                    upstream.provider_type,
+                    model_id,
+                    response.status_code,
+                    path,
                     extra={
                         "status_code": response.status_code,
                         "path": path,
+                        "provider": upstream.provider_type,
+                        "model": model_id,
                         "key_hash": key.hashed_key[:8] + "...",
                         "key_balance": key.balance,
                         "max_cost_for_model": max_cost_for_model,
-                        "upstream_headers": response.headers
-                        if hasattr(response, "headers")
-                        else None,
                     },
                 )
                 return response
 
             return response
 
+        except asyncio.CancelledError:
+            logger.warning(
+                "Client disconnected mid-request, reverting reservation",
+                extra={
+                    "path": path,
+                    "model": model_id,
+                    "key_hash": key.hashed_key[:8] + "...",
+                    "max_cost_for_model": max_cost_for_model,
+                },
+            )
+            await asyncio.shield(
+                revert_pay_for_request(key, session, max_cost_for_model)
+            )
+            raise
+
         except UpstreamError as e:
             logger.warning(
-                f"Upstream {upstream.provider_type} failed: {e}",
-                extra={"retry": i < len(upstreams) - 1},
+                "Upstream %s failed for model=%s: %s",
+                upstream.provider_type,
+                model_id,
+                e,
+                extra={
+                    "provider": upstream.provider_type,
+                    "model": model_id,
+                    "status_code": e.status_code,
+                    "retry": i < len(upstreams) - 1,
+                },
             )
 
             # If this was the last provider
             if i == len(upstreams) - 1:
                 await revert_pay_for_request(key, session, max_cost_for_model)
-                return create_error_response(
-                    "upstream_error", str(e), 502, request=request
-                )
+                return create_upstream_error_response(e, request)
 
             # Otherwise loop continues to next provider
             continue
@@ -387,10 +641,16 @@ async def proxy(
 
 
 async def get_bearer_token_key(
-    headers: dict, path: str, session: AsyncSession, auth: str, min_cost: int = 0
+    headers: dict,
+    path: str,
+    session: AsyncSession,
+    auth: str,
+    min_cost: int = 0,
+    model_id: str = "unknown",
 ) -> ApiKey:
     """Handle bearer token authentication proxy requests."""
-    bearer_key = auth.replace("Bearer ", "") if auth.startswith("Bearer ") else ""
+    parts = auth.split()
+    bearer_key = parts[1] if len(parts) > 1 and parts[0].lower() == "bearer" else ""
     refund_address = headers.get("Refund-LNURL", None)
     key_expiry_time = headers.get("Key-Expiry-Time", None)
 
@@ -454,15 +714,16 @@ async def get_bearer_token_key(
         )
         return key
     except Exception as e:
+        key_preview = bearer_key[:20] + "..." if len(bearer_key) > 20 else bearer_key
         logger.error(
-            "Bearer token validation failed",
+            f"Bearer token validation failed: {type(e).__name__}: {e} path={path} model={model_id!r} min_cost={min_cost} key={key_preview!r}",
             extra={
                 "error": str(e),
                 "error_type": type(e).__name__,
                 "path": path,
-                "bearer_key_preview": bearer_key[:20] + "..."
-                if len(bearer_key) > 20
-                else bearer_key,
+                "model_id": model_id,
+                "min_cost_msat": min_cost,
+                "bearer_key_preview": key_preview,
             },
         )
         raise

routstr/upstream/__init__.py

@@ -10,6 +10,8 @@ from .openai import OpenAIUpstreamProvider
 from .openrouter import OpenRouterUpstreamProvider
 from .perplexity import PerplexityUpstreamProvider
 from .ppqai import PPQAIUpstreamProvider
+from .routstr import RoutstrUpstreamProvider
+from .tinfoil import TinfoilUpstreamProvider
 from .xai import XAIUpstreamProvider
 
 upstream_provider_classes: list[type[BaseUpstreamProvider]] = [
@@ -24,6 +26,8 @@ upstream_provider_classes: list[type[BaseUpstreamProvider]] = [
     OpenRouterUpstreamProvider,
     PerplexityUpstreamProvider,
     PPQAIUpstreamProvider,
+    RoutstrUpstreamProvider,
+    TinfoilUpstreamProvider,
     XAIUpstreamProvider,
 ]
 """List of all upstream classes"""

routstr/upstream/anthropic.py

@@ -13,6 +13,8 @@ class AnthropicUpstreamProvider(BaseUpstreamProvider):
     provider_type = "anthropic"
     default_base_url = "https://api.anthropic.com/v1"
     platform_url = "https://console.anthropic.com/settings/keys"
+    supports_anthropic_messages = True
+    litellm_provider_prefix = "anthropic/"
 
     def __init__(self, api_key: str, provider_fee: float = 1.01):
         super().__init__(
@@ -22,7 +24,7 @@ class AnthropicUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "AnthropicUpstreamProvider":
         return cls(

routstr/upstream/auto_topup.py

@@ -0,0 +1,199 @@
+import asyncio
+import json
+
+from sqlmodel import select
+
+from ..core import get_logger
+from ..core.db import (
+    CashuTransaction,
+    UpstreamProviderRow,
+    create_session,
+    store_cashu_transaction,
+)
+from ..wallet import send_token
+from .routstr import RoutstrUpstreamProvider
+
+logger = get_logger(__name__)
+
+# Check every 60 seconds
+AUTO_TOPUP_INTERVAL_SECONDS = 60
+
+
+async def periodic_auto_topup() -> None:
+    """Background task that monitors Routstr provider balances and auto-tops up when below threshold.
+
+    For each Routstr provider with auto_topup enabled in provider_settings:
+    1. Checks the upstream balance via get_balance()
+    2. If balance < topup_threshold, creates a cashu token from the configured mint
+    3. Sends the token to the upstream provider via topup()
+    """
+    # Wait for initial startup to complete
+    await asyncio.sleep(30)
+    logger.info("Auto top-up worker started")
+
+    while True:
+        try:
+            await _run_auto_topup_cycle()
+        except Exception as e:
+            logger.error(
+                "Auto top-up cycle failed",
+                extra={"error": str(e), "error_type": type(e).__name__},
+            )
+
+        await asyncio.sleep(AUTO_TOPUP_INTERVAL_SECONDS)
+
+
+async def _run_auto_topup_cycle() -> None:
+    """Single cycle: check all eligible providers and top up if needed."""
+    async with create_session() as session:
+        query = select(UpstreamProviderRow).where(
+            UpstreamProviderRow.provider_type == "routstr",
+            UpstreamProviderRow.enabled == True,  # noqa: E712
+        )
+        result = await session.exec(query)
+        providers = result.all()
+
+    for row in providers:
+        try:
+            await _check_and_topup(row)
+        except Exception as e:
+            logger.error(
+                "Auto top-up failed for provider",
+                extra={
+                    "provider_id": row.id,
+                    "base_url": row.base_url,
+                    "error": str(e),
+                    "error_type": type(e).__name__,
+                },
+            )
+
+
+async def _check_and_topup(row: UpstreamProviderRow) -> None:
+    """Check a single provider's balance and top up if below threshold."""
+    # Parse provider settings
+    settings: dict = {}
+    if row.provider_settings:
+        try:
+            settings = json.loads(row.provider_settings)
+        except (json.JSONDecodeError, TypeError):
+            return
+
+    if not settings.get("auto_topup"):
+        return
+
+    threshold = settings.get("topup_threshold")
+    amount = settings.get("topup_amount_limit")
+    mint_url = settings.get("topup_mint_url")
+
+    if not threshold or not amount or not mint_url:
+        logger.warning(
+            "Auto top-up enabled but missing configuration",
+            extra={
+                "provider_id": row.id,
+                "has_threshold": bool(threshold),
+                "has_amount": bool(amount),
+                "has_mint": bool(mint_url),
+            },
+        )
+        return
+
+    if not row.api_key:
+        return
+
+    # Instantiate provider and check balance
+    provider = RoutstrUpstreamProvider.from_db_row(row)
+    if provider is None:
+        return
+    balance = await provider.get_balance()
+
+    if balance is None:
+        logger.warning(
+            "Could not fetch balance for auto top-up",
+            extra={"provider_id": row.id, "base_url": row.base_url},
+        )
+        return
+
+    if balance >= threshold * 1000:
+        return
+
+    # Balance is below threshold - create token and top up
+    logger.info(
+        "Auto top-up triggered",
+        extra={
+            "provider_id": row.id,
+            "balance": balance,
+            "threshold": threshold,
+            "topup_amount": amount,
+            "mint_url": mint_url,
+        },
+    )
+
+    try:
+        token = await send_token(amount, "sat", mint_url)
+    except Exception as e:
+        logger.error(
+            "Failed to create cashu token for auto top-up",
+            extra={
+                "provider_id": row.id,
+                "amount": amount,
+                "mint_url": mint_url,
+                "error": str(e),
+            },
+        )
+        return
+
+    stored = await store_cashu_transaction(
+        token=token,
+        amount=amount,
+        unit="sat",
+        mint_url=mint_url,
+        typ="out",
+        collected=False,
+        source="auto_topup",
+    )
+    if not stored:
+        logger.critical(
+            "Aborting auto top-up because its cashu token could not be persisted",
+            extra={"provider_id": row.id, "mint_url": mint_url},
+        )
+        return
+
+    result = await provider.topup(token)
+
+    if "error" in result:
+        logger.error(
+            "Auto top-up upstream call failed",
+            extra={
+                "provider_id": row.id,
+                "error": result["error"],
+            },
+        )
+    else:
+        async with create_session() as session:
+            transaction = (
+                await session.exec(
+                    select(CashuTransaction).where(
+                        CashuTransaction.token == token,
+                        CashuTransaction.type == "out",
+                        CashuTransaction.source == "auto_topup",
+                    )
+                )
+            ).first()
+            if transaction is None:
+                logger.critical(
+                    "Completed auto top-up transaction is missing from the database",
+                    extra={"provider_id": row.id, "mint_url": mint_url},
+                )
+            else:
+                transaction.collected = True
+                session.add(transaction)
+                await session.commit()
+
+        logger.info(
+            "Auto top-up completed successfully",
+            extra={
+                "provider_id": row.id,
+                "amount": amount,
+                "new_balance_approx": balance + amount,
+            },
+        )

routstr/upstream/azure.py

@@ -4,6 +4,7 @@ from .base import BaseUpstreamProvider
 
 if TYPE_CHECKING:
     from ..core.db import UpstreamProviderRow
+    from ..payment.models import Model
 
 
 class AzureUpstreamProvider(BaseUpstreamProvider):
@@ -12,6 +13,7 @@ class AzureUpstreamProvider(BaseUpstreamProvider):
     provider_type = "azure"
     default_base_url = None
     platform_url = "https://portal.azure.com/"
+    litellm_provider_prefix = "azure/"
 
     def __init__(
         self,
@@ -36,7 +38,7 @@ class AzureUpstreamProvider(BaseUpstreamProvider):
         self.api_version = api_version
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "AzureUpstreamProvider | None":
         if not provider_row.api_version:
@@ -58,19 +60,52 @@ class AzureUpstreamProvider(BaseUpstreamProvider):
             "platform_url": cls.platform_url,
         }
 
+    def prepare_headers(self, request_headers: dict) -> dict:
+        """Prepare headers for Azure OpenAI, adding api-key."""
+        headers = super().prepare_headers(request_headers)
+        if self.api_key:
+            headers["api-key"] = self.api_key
+            headers.pop("Authorization", None)
+            headers.pop("authorization", None)
+        return headers
+
     def prepare_params(
         self, path: str, query_params: Mapping[str, str] | None
     ) -> Mapping[str, str]:
-        """Prepare query parameters for Azure OpenAI, adding API version.
-
-        Args:
-            path: Request path
-            query_params: Original query parameters from the client
-
-        Returns:
-            Query parameters dict with Azure API version added for chat completions
-        """
+        """Prepare query parameters for Azure OpenAI, adding API version."""
         params = dict(query_params or {})
-        if path.endswith("chat/completions"):
-            params["api-version"] = self.api_version
+        version = (self.api_version or "").replace("\ufeff", "").strip()
+        if not version or version.lower() == "v1":
+            version = "2024-02-15-preview"
+        params["api-version"] = version
         return params
+
+    def normalize_request_path(
+        self, path: str, model_obj: "Model | None" = None
+    ) -> str:
+        """Build Azure deployment-specific request path."""
+        clean_path = super().normalize_request_path(path, model_obj).lstrip("/")
+        if model_obj is None:
+            return clean_path
+
+        deployment_id = getattr(
+            model_obj, "canonical_slug", None
+        ) or self.transform_model_name(model_obj.id)
+        deployment_id = deployment_id.split("/")[-1]
+        return f"openai/deployments/{deployment_id}/{clean_path}"
+
+    def get_request_base_url(
+        self, path: str, model_obj: "Model | None" = None
+    ) -> str:
+        """Use endpoint root, stripping accidental /openai/v1 suffix if present."""
+        base_url = self.base_url.rstrip("/")
+        marker = "/openai/v1"
+        if marker in base_url:
+            base_url = base_url.split(marker, 1)[0].rstrip("/")
+        return base_url
+
+    def transform_model_name(self, model_id: str) -> str:
+        """Extract deployment name from model ID."""
+        if "/" in model_id:
+            return model_id.split("/")[-1]
+        return model_id

routstr/upstream/cache_breakpoints.py

@@ -0,0 +1,157 @@
+"""Inject explicit prompt-cache breakpoints into OpenAI-shaped requests.
+
+Some upstreams cache *explicitly*: the request must carry
+``cache_control: {"type": "ephemeral"}`` markers on the content blocks that
+should be cached. Two model families use this identical wire format:
+
+* **Anthropic Claude** — direct or via OpenRouter's ``anthropic/*`` models.
+* **Alibaba's explicit-cache models on OpenRouter** — ``qwen/qwen3-max``,
+  ``qwen/qwen-plus``, ``qwen/qwen3.6-plus``, ``qwen/qwen3-coder-plus``,
+  ``qwen/qwen3-coder-flash`` and ``deepseek/deepseek-v3.2`` — which OpenRouter
+  documents as using "the same syntax as Anthropic explicit caching".
+
+Every other provider routstr proxies (OpenAI, Azure, xAI/Grok, Groq, Moonshot,
+default DeepSeek, Gemini implicit, Fireworks) caches *automatically* and needs
+no markers — they are left untouched.
+
+A client that doesn't know it is talking to one of these models *through*
+routstr (e.g. an OpenAI-compatible coding agent pointed at a routstr URL) never
+emits the markers — it only adds them when it recognises the provider as
+OpenRouter. So caching silently never engages over routstr even though the same
+client caches fine talking to OpenRouter directly.
+
+This module restores caching by stamping the standard breakpoints onto the
+forwarded body — the system prompt, the last tool, and the last conversation
+message (the format allows up to four; we use three, matching the common
+agent convention) — but only when the client supplied none of its own, so
+explicit client control always wins. The caller is responsible for only
+applying this toward an upstream that accepts the markers (OpenRouter /
+Anthropic), so they never leak to a provider that would reject them.
+"""
+
+from __future__ import annotations
+
+from typing import Any
+
+# The single ephemeral marker stamped onto each chosen breakpoint. A 5-minute
+# TTL (the default for ``ephemeral``) — deliberately not the 1h tier, which
+# carries a higher cache-write premium and should stay opt-in.
+EPHEMERAL_CACHE_CONTROL: dict[str, str] = {"type": "ephemeral"}
+
+# Alibaba's explicit-cache models on OpenRouter. Matched as substrings of the
+# model id (any spelling routstr carries). Snapshot endpoints that OpenRouter
+# documents as *not* supporting explicit caching (e.g. ``qwen3.5-plus-02-15``)
+# are different families and deliberately absent from this list.
+_ALIBABA_EXPLICIT_CACHE_SLUGS: tuple[str, ...] = (
+    "qwen3-max",
+    "qwen-plus",
+    "qwen3.6-plus",
+    "qwen3-coder-plus",
+    "qwen3-coder-flash",
+    "deepseek-v3.2",
+)
+
+
+def is_explicit_cache_model(model_id: str | None, *fallbacks: str | None) -> bool:
+    """True when the target model uses the explicit ``cache_control`` dialect.
+
+    Covers the Claude family (broadly — every Claude model supports it) and
+    Alibaba's documented explicit-cache models, across the id spellings routstr
+    carries: the OpenRouter id (``anthropic/claude-...``, ``qwen/qwen3-max``),
+    the bare upstream id, and any forwarded/canonical alias.
+    """
+    for candidate in (model_id, *fallbacks):
+        if not candidate:
+            continue
+        lowered = candidate.lower()
+        if "claude" in lowered or "anthropic/" in lowered:
+            return True
+        if any(slug in lowered for slug in _ALIBABA_EXPLICIT_CACHE_SLUGS):
+            return True
+    return False
+
+
+def _has_cache_control(obj: Any) -> bool:
+    """Recursively detect any client-supplied ``cache_control`` marker."""
+    if isinstance(obj, dict):
+        if "cache_control" in obj:
+            return True
+        return any(_has_cache_control(v) for v in obj.values())
+    if isinstance(obj, list):
+        return any(_has_cache_control(v) for v in obj)
+    return False
+
+
+def body_has_cache_control(data: dict) -> bool:
+    """True when the request already carries cache_control on messages/tools."""
+    return _has_cache_control(data.get("messages")) or _has_cache_control(
+        data.get("tools")
+    )
+
+
+def _stamp_text_content(message: dict) -> bool:
+    """Add the ephemeral marker to a message's last text block.
+
+    A string content is promoted to the array form Anthropic requires for
+    cache markers; an existing array gets the marker on its last text part.
+    Returns True when a marker was placed.
+    """
+    content = message.get("content")
+    if isinstance(content, str):
+        if not content:
+            return False
+        message["content"] = [
+            {
+                "type": "text",
+                "text": content,
+                "cache_control": dict(EPHEMERAL_CACHE_CONTROL),
+            }
+        ]
+        return True
+    if isinstance(content, list):
+        for part in reversed(content):
+            if isinstance(part, dict) and part.get("type") == "text":
+                part["cache_control"] = dict(EPHEMERAL_CACHE_CONTROL)
+                return True
+    return False
+
+
+def _stamp_system_prompt(messages: list) -> None:
+    for message in messages:
+        if isinstance(message, dict) and message.get("role") in (
+            "system",
+            "developer",
+        ):
+            _stamp_text_content(message)
+            return
+
+
+def _stamp_last_tool(tools: Any) -> None:
+    if isinstance(tools, list) and tools and isinstance(tools[-1], dict):
+        tools[-1]["cache_control"] = dict(EPHEMERAL_CACHE_CONTROL)
+
+
+def _stamp_last_conversation_message(messages: list) -> None:
+    for message in reversed(messages):
+        if isinstance(message, dict) and message.get("role") in ("user", "assistant"):
+            if _stamp_text_content(message):
+                return
+
+
+def inject_anthropic_cache_breakpoints(data: dict) -> bool:
+    """Stamp ephemeral cache breakpoints onto an OpenAI-shaped chat body.
+
+    Mutates ``data`` in place (the established convention in
+    ``prepare_request_body``) and returns True when anything changed. No-ops
+    when the body isn't chat-shaped or the client already set cache_control.
+    """
+    messages = data.get("messages")
+    if not isinstance(messages, list) or not messages:
+        return False
+    if body_has_cache_control(data):
+        return False
+
+    _stamp_system_prompt(messages)
+    _stamp_last_tool(data.get("tools"))
+    _stamp_last_conversation_message(messages)
+    return True

routstr/upstream/clients/gemini.py

@@ -26,14 +26,14 @@ class GeminiClient(BaseAPIClient):
         max_tokens: int | None = None,
         **kwargs: Any,
     ) -> dict[str, Any]:
-        from openai import NOT_GIVEN
+        from openai import omit
 
         response = await self.client.chat.completions.create(
             model=model,
             messages=messages,  # type: ignore
-            temperature=temperature if temperature is not None else NOT_GIVEN,
-            max_tokens=max_tokens if max_tokens is not None else NOT_GIVEN,
-            top_p=kwargs.get("top_p", NOT_GIVEN),
+            temperature=temperature if temperature is not None else omit,
+            max_tokens=max_tokens if max_tokens is not None else omit,
+            top_p=kwargs.get("top_p", omit),
         )
         return response.model_dump()
 
@@ -45,7 +45,7 @@ class GeminiClient(BaseAPIClient):
         max_tokens: int | None = None,
         **kwargs: Any,
     ) -> AsyncGenerator[dict[str, Any], None]:
-        from openai import NOT_GIVEN
+        from openai import omit
 
         usage_callback = kwargs.get("usage_callback")
         completion_callback = kwargs.get("completion_callback")
@@ -55,9 +55,9 @@ class GeminiClient(BaseAPIClient):
             messages=messages,  # type: ignore
             stream=True,
             stream_options={"include_usage": True},
-            temperature=temperature if temperature is not None else NOT_GIVEN,
-            max_tokens=max_tokens if max_tokens is not None else NOT_GIVEN,
-            top_p=kwargs.get("top_p", NOT_GIVEN),
+            temperature=temperature if temperature is not None else omit,
+            max_tokens=max_tokens if max_tokens is not None else omit,
+            top_p=kwargs.get("top_p", omit),
         )
 
         final_usage = None

routstr/upstream/count_tokens.py

@@ -0,0 +1,108 @@
+"""Local handling of Anthropic ``/v1/messages/count_tokens`` for upstreams
+that do not natively expose the endpoint.
+
+Most non-Anthropic upstreams (OpenAI-compat, Gemini OpenAI-compat,
+OpenRouter chat-completions, generic providers) return 400/404 when asked
+to ``POST /messages/count_tokens``. Claude Code and other Anthropic SDK
+clients call this endpoint before each turn to size context windows and
+trigger compaction, so a failure breaks the whole chat.
+
+We answer locally. ``litellm.token_counter`` understands the Anthropic
+message shape and the per-model tokenizers, so we prefer it. If it raises
+(unknown model, encoding lookup failure, ...), we fall back to the
+project's own ``estimate_tokens`` heuristic, which is always defined and
+never raises.
+"""
+
+from __future__ import annotations
+
+import json
+from typing import Any
+
+import litellm
+from fastapi.responses import Response
+
+from ..core import get_logger
+from ..payment.helpers import estimate_tokens
+from ..payment.models import Model
+
+logger = get_logger(__name__)
+
+
+def _parse_request_body(request_body: bytes | None) -> dict[str, Any]:
+    if not request_body:
+        return {}
+    try:
+        parsed = json.loads(request_body)
+    except (ValueError, TypeError):
+        return {}
+    return parsed if isinstance(parsed, dict) else {}
+
+
+def _count_with_litellm(model: str, body: dict[str, Any]) -> int:
+    messages = body.get("messages")
+    if not isinstance(messages, list):
+        messages = []
+
+    system = body.get("system")
+    if isinstance(system, str) and system:
+        messages = [{"role": "system", "content": system}, *messages]
+    elif isinstance(system, list):
+        text = "".join(
+            block.get("text", "")
+            for block in system
+            if isinstance(block, dict) and block.get("type") == "text"
+        )
+        if text:
+            messages = [{"role": "system", "content": text}, *messages]
+
+    tools = body.get("tools") if isinstance(body.get("tools"), list) else None
+
+    return int(
+        litellm.token_counter(
+            model=model,
+            messages=messages,
+            tools=tools,
+        )
+    )
+
+
+def count_tokens_locally(
+    request_body: bytes | None,
+    model_obj: Model | None,
+) -> Response:
+    """Return an Anthropic-compatible count_tokens response without
+    touching the upstream. Always returns 200; never raises."""
+    body = _parse_request_body(request_body)
+
+    model_name = ""
+    if model_obj is not None:
+        model_name = model_obj.forwarded_model_id or model_obj.id or ""
+    if not model_name:
+        body_model = body.get("model")
+        if isinstance(body_model, str):
+            model_name = body_model
+
+    input_tokens: int
+    try:
+        input_tokens = _count_with_litellm(model_name, body)
+    except Exception as exc:
+        messages = body.get("messages")
+        fallback_messages = messages if isinstance(messages, list) else []
+        input_tokens = estimate_tokens(fallback_messages)
+        logger.debug(
+            "litellm token_counter failed; using local estimator",
+            extra={
+                "model": model_name,
+                "error": str(exc),
+                "error_type": type(exc).__name__,
+                "estimated_tokens": input_tokens,
+            },
+        )
+
+    payload = {"input_tokens": max(0, int(input_tokens))}
+    return Response(
+        content=json.dumps(payload).encode(),
+        status_code=200,
+        media_type="application/json",
+    )

routstr/upstream/deepseek_v4_pricing_shim.py

@@ -0,0 +1,73 @@
+"""TEMPORARY: local DeepSeek V4 pricing shim.
+
+litellm's bundled cost map does not yet ship ``deepseek-v4-flash`` /
+``deepseek-v4-pro``. Without an entry, ``backfill_cache_pricing`` cannot find a
+``cache_read_input_token_cost`` and cache reads fall back to the full input
+rate — a large overcharge on cache hits (DeepSeek V4 hits are ~0.008-0.02x
+input, i.e. cached tokens cost 50-120x less than regular input).
+
+This module injects the missing entries into ``litellm.model_cost`` at startup
+so the existing backfill path resolves them. Rates mirror the canonical
+``deepseek`` provider entries now in litellm's ``model_prices`` map
+(``input_cost_per_token`` is the cache-*miss* rate;
+``cache_read_input_token_cost`` is the cache-*hit* rate), sourced from
+https://api-docs.deepseek.com/quick_start/pricing via
+https://github.com/BerriAI/litellm/pull/26380 (issue
+https://github.com/BerriAI/litellm/issues/30430).
+
+=== REMOVAL (once litellm ships these models) ===
+Delete this file and the single ``register_deepseek_v4_pricing()`` call in
+``routstr/core/main.py``. Nothing else depends on it. Entries are only added
+when absent, so a stale shim is harmless after upstream lands — but remove it.
+"""
+
+import litellm
+
+from ..core import get_logger
+
+logger = get_logger(__name__)
+
+# USD per token. Mirrors the canonical ``deepseek`` provider entries in
+# litellm's model_prices map (source: DeepSeek API pricing docs). Keep these in
+# sync with ``litellm.model_cost["deepseek/deepseek-v4-*"]``.
+_DEEPSEEK_V4_RATES: dict[str, dict[str, float]] = {
+    "deepseek-v4-flash": {
+        "input_cost_per_token": 1.4e-07,
+        "output_cost_per_token": 2.8e-07,
+        "cache_read_input_token_cost": 2.8e-09,
+        "cache_creation_input_token_cost": 0.0,
+        "input_cost_per_token_cache_hit": 2.8e-09,
+    },
+    "deepseek-v4-pro": {
+        "input_cost_per_token": 4.35e-07,
+        "output_cost_per_token": 8.7e-07,
+        "cache_read_input_token_cost": 3.625e-09,
+        "cache_creation_input_token_cost": 0.0,
+        "input_cost_per_token_cache_hit": 3.625e-09,
+    },
+}
+
+
+def register_deepseek_v4_pricing() -> None:
+    """Inject DeepSeek V4 pricing into ``litellm.model_cost`` if absent.
+
+    Idempotent and non-destructive: a key already present in the cost map
+    (e.g. once litellm ships it) is left untouched. Registers both the bare
+    (``deepseek-v4-flash``) and prefixed (``deepseek/deepseek-v4-flash``)
+    spellings since ``backfill_cache_pricing`` tries both.
+    """
+    added = []
+    for bare, rates in _DEEPSEEK_V4_RATES.items():
+        for key in (bare, f"deepseek/{bare}"):
+            if key in litellm.model_cost:
+                continue
+            entry: dict[str, object] = dict(rates)
+            entry["litellm_provider"] = "deepseek"
+            entry["mode"] = "chat"
+            litellm.model_cost[key] = entry
+            added.append(key)
+    if added:
+        logger.info(
+            "Registered temporary DeepSeek V4 pricing shim",
+            extra={"models": added},
+        )

routstr/upstream/fireworks.py

@@ -12,6 +12,7 @@ class FireworksUpstreamProvider(BaseUpstreamProvider):
     provider_type = "fireworks"
     default_base_url = "https://api.fireworks.ai/inference/v1"
     platform_url = "https://app.fireworks.ai/settings/users/api-keys"
+    litellm_provider_prefix = "fireworks_ai/"
 
     def __init__(self, api_key: str, provider_fee: float = 1.01):
         super().__init__(
@@ -19,7 +20,7 @@ class FireworksUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "FireworksUpstreamProvider":
         return cls(

routstr/upstream/gemini.py

@@ -1,17 +1,13 @@
 from __future__ import annotations
 
-import json
-from collections.abc import AsyncGenerator
 from typing import TYPE_CHECKING, Any
 
-from fastapi import Request
-from fastapi.responses import Response, StreamingResponse
-
+from . import gemini_messages
 from .base import BaseUpstreamProvider
 from .clients.gemini import GeminiClient
 
 if TYPE_CHECKING:
-    from ..core.db import ApiKey, AsyncSession, UpstreamProviderRow
+    from ..core.db import UpstreamProviderRow
     from ..payment.models import Model
 
 from ..core.logging import get_logger
@@ -20,9 +16,19 @@ logger = get_logger(__name__)
 
 
 class GeminiUpstreamProvider(BaseUpstreamProvider):
+    """Gemini provider — proxies through Gemini's OpenAI-compat surface.
+
+    The chat-completions, embeddings, and models paths all flow through
+    :meth:`BaseUpstreamProvider.forward_request`; we only override
+    ``get_request_base_url`` to redirect to ``{base}/openai/...`` and
+    ``_dispatch_anthropic_messages`` to inject thought-signatures on the
+    /v1/messages path (see :mod:`gemini_messages` for that rationale).
+    """
+
     provider_type = "gemini"
     default_base_url = "https://generativelanguage.googleapis.com/v1beta"
     platform_url = "https://aistudio.google.com/app/apikey"
+    litellm_provider_prefix = "gemini/"
 
     def __init__(
         self,
@@ -39,13 +45,13 @@ class GeminiUpstreamProvider(BaseUpstreamProvider):
 
     @property
     def client(self) -> GeminiClient:
-        """Get or create the Gemini API client."""
+        """Get or create the Gemini API client (used for the models listing)."""
         if self._client is None:
             self._client = GeminiClient(api_key=self.api_key)
         return self._client
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "GeminiUpstreamProvider":
         return cls(
@@ -65,240 +71,66 @@ class GeminiUpstreamProvider(BaseUpstreamProvider):
         }
 
     def transform_model_name(self, model_id: str) -> str:
-        return model_id.removeprefix("gemini/")
+        """Reduce a routstr model id to the bare upstream Gemini name.
 
-    async def forward_request(
+        Gemini's OpenAI-compat surface expects the literal model id
+        (e.g. ``gemini-3.1-flash-lite-preview``) — no ``gemini/`` provider
+        prefix and no ``google/`` vendor sub-prefix. Take the last path
+        segment so we tolerate any of:
+
+            ``gemini-2.0-flash``
+            ``gemini/gemini-2.0-flash``
+            ``gemini/google/gemini-3.1-flash-lite-preview``
+        """
+        return model_id.rsplit("/", 1)[-1]
+
+    @property
+    def compat_base_url(self) -> str:
+        """Gemini's OpenAI-compat surface, regardless of what's stored.
+
+        Stored ``base_url`` may be ``.../v1beta`` (the native Gemini API
+        root) or ``.../v1beta/openai`` (already pointed at the compat
+        surface). Normalize to the latter.
+        """
+        return self.base_url.rstrip("/").removesuffix("/openai") + "/openai"
+
+    def get_request_base_url(
+        self, path: str, model_obj: "Model | None" = None
+    ) -> str:
+        """Route every proxied request to the OpenAI-compat surface.
+
+        Required because the stored ``base_url`` typically points at the
+        native Gemini API (``/v1beta``), but :meth:`forward_request`
+        forwards OpenAI-shaped paths (``/chat/completions``,
+        ``/embeddings``, ``/models``) which only exist under the
+        ``/openai`` subtree.
+        """
+        return self.compat_base_url
+
+    async def _dispatch_anthropic_messages(
         self,
-        request: Request,
-        path: str,
-        headers: dict,
         request_body: bytes | None,
-        key: ApiKey,
-        max_cost_for_model: int,
-        session: AsyncSession,
-        model_obj: Model,
-    ) -> Response | StreamingResponse:
-        # Remove provider prefix from model ID for Gemini API
-        if "/" in model_obj.id:
-            model_obj.id = model_obj.id.split("/", 1)[1]
+        model_obj: "Model",
+        *,
+        log_extra: dict[str, Any] | None = None,
+    ) -> tuple[bool, Any, str | None]:
+        """Dispatch /v1/messages via the gemini-specific httpx path.
 
-        if not path.startswith("chat/completions"):
-            return await super().forward_request(
-                request,
-                path,
-                headers,
-                request_body,
-                key,
-                max_cost_for_model,
-                session,
-                model_obj,
-            )
-
-        if not request_body:
-            return await super().forward_request(
-                request,
-                path,
-                headers,
-                request_body,
-                key,
-                max_cost_for_model,
-                session,
-                model_obj,
-            )
-
-        try:
-            openai_data = json.loads(request_body)
-            messages = openai_data.get("messages", [])
-            temperature = openai_data.get("temperature")
-            max_tokens = openai_data.get("max_tokens")
-            top_p = openai_data.get("top_p")
-            is_streaming = openai_data.get("stream", False)
-
-            logger.info(
-                "Processing Gemini request with client abstraction",
-                extra={
-                    "model": model_obj.id,
-                    "is_streaming": is_streaming,
-                    "message_count": len(messages),
-                    "key_hash": key.hashed_key[:8] + "...",
-                },
-            )
-
-            if is_streaming:
-                final_usage_data: dict | None = None
-
-                def usage_callback(usage_data: dict[str, Any]) -> None:
-                    """Callback to capture usage data during streaming"""
-                    nonlocal final_usage_data
-                    final_usage_data = usage_data
-
-                async def completion_callback(
-                    model: str, usage_data: dict[str, Any] | None
-                ) -> None:
-                    """Callback to handle payment when streaming completes"""
-                    nonlocal final_usage_data
-                    if usage_data:
-                        final_usage_data = usage_data
-
-                    payment_data = {
-                        "model": model,
-                        "usage": final_usage_data,
-                    }
-
-                    from ..auth import adjust_payment_for_tokens
-                    from ..core.db import create_session
-
-                    async with create_session() as new_session:
-                        fresh_key = await new_session.get(key.__class__, key.hashed_key)
-                        if fresh_key:
-                            try:
-                                cost_data = await adjust_payment_for_tokens(
-                                    fresh_key,
-                                    payment_data,
-                                    new_session,
-                                    max_cost_for_model,
-                                )
-
-                                logger.info(
-                                    "Gemini streaming payment finalized",
-                                    extra={
-                                        "cost_data": cost_data,
-                                        "usage_data": final_usage_data,
-                                        "key_hash": key.hashed_key[:8] + "...",
-                                    },
-                                )
-                            except Exception as cost_error:
-                                logger.error(
-                                    "Error finalizing Gemini streaming payment",
-                                    extra={
-                                        "error": str(cost_error),
-                                        "key_hash": key.hashed_key[:8] + "...",
-                                    },
-                                )
-
-                response_generator = self.client.generate_content_stream(
-                    model=model_obj.id,
-                    messages=messages,
-                    temperature=temperature,
-                    max_tokens=max_tokens,
-                    top_p=top_p,
-                    usage_callback=usage_callback,
-                    completion_callback=completion_callback,
-                )
-
-                async def stream_with_cost() -> AsyncGenerator[bytes, None]:
-                    payment_finalized = False
-
-                    async def finalize_payment() -> None:
-                        nonlocal payment_finalized
-                        if payment_finalized:
-                            return
-                        from ..auth import adjust_payment_for_tokens
-                        from ..core.db import create_session
-
-                        async with create_session() as new_session:
-                            fresh_key = await new_session.get(
-                                key.__class__, key.hashed_key
-                            )
-                            if fresh_key:
-                                try:
-                                    await adjust_payment_for_tokens(
-                                        fresh_key,
-                                        {
-                                            "model": model_obj.id,
-                                            "usage": final_usage_data,
-                                        },
-                                        new_session,
-                                        max_cost_for_model,
-                                    )
-                                    payment_finalized = True
-                                except Exception as cost_error:
-                                    logger.error(
-                                        "Error finalizing Gemini streaming payment in fallback",
-                                        extra={
-                                            "error": str(cost_error),
-                                            "key_hash": key.hashed_key[:8] + "...",
-                                        },
-                                    )
-
-                    try:
-                        async for chunk in response_generator:
-                            sse_data = f"data: {json.dumps(chunk)}\n\n"
-                            yield sse_data.encode()
-                    except Exception as e:
-                        logger.error(
-                            "Error in Gemini streaming response",
-                            extra={
-                                "error": str(e),
-                                "error_type": type(e).__name__,
-                                "key_hash": key.hashed_key[:8] + "...",
-                            },
-                        )
-                        raise
-                    finally:
-                        if not payment_finalized:
-                            await finalize_payment()
-
-                return StreamingResponse(
-                    stream_with_cost(),
-                    media_type="text/event-stream",
-                    headers={"Cache-Control": "no-cache", "Connection": "keep-alive"},
-                )
-
-            else:
-                openai_format_response = await self.client.generate_content(
-                    model=model_obj.id,
-                    messages=messages,
-                    temperature=temperature,
-                    max_tokens=max_tokens,
-                    top_p=top_p,
-                )
-
-                from ..auth import adjust_payment_for_tokens
-
-                cost_data = await adjust_payment_for_tokens(
-                    key, openai_format_response, session, max_cost_for_model
-                )
-                openai_format_response["cost"] = cost_data
-
-                logger.info(
-                    "Gemini non-streaming payment completed",
-                    extra={
-                        "cost_data": cost_data,
-                        "model": model_obj.id,
-                        "key_hash": key.hashed_key[:8] + "...",
-                    },
-                )
-
-                return Response(
-                    content=json.dumps(openai_format_response),
-                    media_type="application/json",
-                    headers={"Cache-Control": "no-cache"},
-                )
-
-        except Exception as e:
-            logger.error(
-                "Error in Gemini forward_request",
-                extra={
-                    "error": str(e),
-                    "error_type": type(e).__name__,
-                    "path": path,
-                    "key_hash": key.hashed_key[:8] + "...",
-                },
-            )
-            return await super().forward_request(
-                request,
-                path,
-                headers,
-                request_body,
-                key,
-                max_cost_for_model,
-                session,
-                model_obj,
-            )
+        See :mod:`routstr.upstream.gemini_messages` for the full rationale
+        (thought-signature injection, why litellm + the openai SDK can't
+        carry the required ``extra_content`` field).
+        """
+        return await gemini_messages.dispatch_gemini_messages(
+            request_body=request_body,
+            model_obj=model_obj,
+            base_url=self.compat_base_url,
+            api_key=self.api_key,
+            transform_model_name=self.transform_model_name,
+            log_extra=log_extra,
+        )
 
     async def _fetch_provider_models(self) -> dict:
-        """Fetch models from Gemini API."""
+        """Fetch models from Gemini API via the OpenAI-compat client."""
         try:
             models_data = await self.client.list_models()
 

routstr/upstream/gemini_messages.py

@@ -0,0 +1,480 @@
+"""Custom /v1/messages dispatcher for Gemini's OpenAI-compat endpoint.
+
+Why this exists
+---------------
+
+Gemini 2.5 / 3 thinking models reject inbound ``functionCall`` parts that
+lack a ``thought_signature`` field once any prior turn in the conversation
+contains a function call. Anthropic-Messages clients (Claude Code etc.)
+have no concept of thought signatures, so multi-turn tool conversations
+fail with::
+
+    Function call is missing a thought_signature in functionCall parts.
+
+Google's published escape hatch (https://ai.google.dev/gemini-api/docs/
+thought-signatures, FAQ #1) is the dummy signature
+``"skip_thought_signature_validator"`` placed at
+``tool_calls[i].extra_content.google.thought_signature`` for every tool
+call in the request. The hatch is documented specifically for
+"transferring a trace from a different model that does not include thought
+signatures" — exactly our case.
+
+Why we can't reach the wire via litellm
+---------------------------------------
+
+``litellm.anthropic.messages.acreate`` flows through the openai SDK, whose
+pydantic ``ChatCompletionMessageToolCall`` model silently drops unknown
+fields like ``extra_content``. Litellm has no openai-compat translator
+that emits ``extra_content.google.thought_signature``. So we bypass both
+litellm and the openai SDK at the transport layer.
+
+Pipeline
+--------
+
+1. Translate Anthropic body → OpenAI body via litellm's
+   ``AnthropicAdapter`` (the same translator
+   ``litellm.anthropic.messages.acreate`` uses internally).
+2. Inject ``extra_content.google.thought_signature`` on every
+   ``tool_calls[]`` entry.
+3. Set ``reasoning_effort="none"`` to disable Gemini's thinking pass.
+4. POST directly to ``{base_url}/chat/completions`` with ``stream=true``
+   via ``httpx`` (preserves arbitrary fields verbatim).
+5. Translate OpenAI streaming chunks → Anthropic SSE events.
+"""
+
+from __future__ import annotations
+
+import json
+import uuid
+from collections.abc import AsyncGenerator, AsyncIterator
+from typing import Any, Callable
+
+import httpx
+
+from ..core import get_logger
+from ..core.exceptions import UpstreamError
+from ..payment.models import Model
+from .messages_dispatch import (
+    ANTHROPIC_ONLY_FIELDS,
+    aggregate_anthropic_events_to_message,
+)
+
+logger = get_logger(__name__)
+
+DUMMY_THOUGHT_SIGNATURE = "skip_thought_signature_validator"
+
+# Mapping: OpenAI finish_reason → Anthropic stop_reason
+_FINISH_TO_STOP = {
+    "stop": "end_turn",
+    "length": "max_tokens",
+    "tool_calls": "tool_use",
+    "function_call": "tool_use",
+    "content_filter": "refusal",
+}
+
+
+def inject_thought_signatures(messages: list[dict]) -> None:
+    """Add ``extra_content.google.thought_signature`` to every tool_call.
+
+    Mutates ``messages`` in place. Idempotent: existing signatures are not
+    overwritten.
+    """
+    for msg in messages:
+        tool_calls = msg.get("tool_calls")
+        if not isinstance(tool_calls, list):
+            continue
+        for tc in tool_calls:
+            if not isinstance(tc, dict):
+                continue
+            extra = tc.get("extra_content")
+            if not isinstance(extra, dict):
+                extra = tc["extra_content"] = {}
+            google_cfg = extra.get("google")
+            if not isinstance(google_cfg, dict):
+                google_cfg = extra["google"] = {}
+            google_cfg.setdefault("thought_signature", DUMMY_THOUGHT_SIGNATURE)
+
+
+def _translate_anthropic_to_openai(body: dict, model: str) -> dict:
+    """Use litellm's translator to convert an Anthropic /messages body to
+    OpenAI /chat/completions kwargs.
+
+    Imported lazily because the litellm internal path is heavy and not
+    needed for any other code path in routstr.
+    """
+    from litellm.llms.anthropic.experimental_pass_through.adapters.transformation import (  # noqa: E501
+        AnthropicAdapter,
+    )
+
+    kwargs = {"model": model, **body}
+    translated = AnthropicAdapter().translate_completion_input_params(kwargs)
+    if translated is None:
+        raise UpstreamError(
+            "Failed to translate Anthropic body to OpenAI format",
+            status_code=500,
+        )
+    return dict(translated)
+
+
+def _sse_event(event_type: str, payload: dict) -> bytes:
+    return f"event: {event_type}\ndata: {json.dumps(payload)}\n\n".encode()
+
+
+async def _openai_chunks_to_anthropic_events(
+    line_iter: AsyncIterator[str], requested_model: str | None
+) -> AsyncGenerator[bytes, None]:
+    """Translate an OpenAI chat-completions SSE byte stream into the
+    Anthropic-Messages SSE event sequence.
+
+    Maintains per-chunk state across:
+
+    * one optional text content block (lazy-opened on first text delta)
+    * any number of tool_use blocks indexed by openai's ``delta.tool_calls[].index``
+    * final ``stop_reason`` / ``usage`` carried out via ``message_delta`` /
+      ``message_stop``
+    """
+    msg_id = f"msg_{uuid.uuid4().hex[:24]}"
+    started = False
+    text_block_idx: int | None = None
+    tool_block_indices: dict[int, int] = {}
+    next_block_idx = 0
+    final_finish_reason: str | None = None
+    final_usage: dict[str, int] = {"input_tokens": 0, "output_tokens": 0}
+
+    def open_text_block() -> bytes:
+        nonlocal text_block_idx, next_block_idx
+        text_block_idx = next_block_idx
+        next_block_idx += 1
+        return _sse_event(
+            "content_block_start",
+            {
+                "type": "content_block_start",
+                "index": text_block_idx,
+                "content_block": {"type": "text", "text": ""},
+            },
+        )
+
+    def open_tool_block(delta_idx: int, tc: dict) -> bytes:
+        nonlocal next_block_idx
+        block_idx = next_block_idx
+        next_block_idx += 1
+        tool_block_indices[delta_idx] = block_idx
+        fn = tc.get("function") or {}
+        return _sse_event(
+            "content_block_start",
+            {
+                "type": "content_block_start",
+                "index": block_idx,
+                "content_block": {
+                    "type": "tool_use",
+                    "id": tc.get("id") or f"toolu_{uuid.uuid4().hex[:24]}",
+                    "name": fn.get("name") or "",
+                    "input": {},
+                },
+            },
+        )
+
+    def close_block(idx: int) -> bytes:
+        return _sse_event(
+            "content_block_stop",
+            {"type": "content_block_stop", "index": idx},
+        )
+
+    async for raw_line in line_iter:
+        line = raw_line.strip()
+        if not line:
+            continue
+        if not line.startswith("data:"):
+            continue
+        payload = line[5:].lstrip()
+        if not payload or payload == "[DONE]":
+            continue
+        try:
+            chunk = json.loads(payload)
+        except json.JSONDecodeError:
+            continue
+        if not isinstance(chunk, dict):
+            continue
+
+        if not started:
+            started = True
+            yield _sse_event(
+                "message_start",
+                {
+                    "type": "message_start",
+                    "message": {
+                        "id": chunk.get("id") or msg_id,
+                        "type": "message",
+                        "role": "assistant",
+                        "model": requested_model or chunk.get("model") or "",
+                        "content": [],
+                        "stop_reason": None,
+                        "stop_sequence": None,
+                        "usage": {
+                            "input_tokens": 0,
+                            "output_tokens": 0,
+                        },
+                    },
+                },
+            )
+
+        usage = chunk.get("usage")
+        if isinstance(usage, dict):
+            in_tok = usage.get("prompt_tokens") or usage.get("input_tokens") or 0
+            out_tok = usage.get("completion_tokens") or usage.get("output_tokens") or 0
+            if in_tok:
+                final_usage["input_tokens"] = int(in_tok)
+            if out_tok:
+                final_usage["output_tokens"] = int(out_tok)
+
+        choices = chunk.get("choices") or []
+        if not choices:
+            continue
+        choice = choices[0] if isinstance(choices[0], dict) else {}
+        delta = choice.get("delta") or {}
+        if not isinstance(delta, dict):
+            delta = {}
+
+        # Text delta
+        text = delta.get("content")
+        if isinstance(text, str) and text:
+            if text_block_idx is None:
+                yield open_text_block()
+            yield _sse_event(
+                "content_block_delta",
+                {
+                    "type": "content_block_delta",
+                    "index": text_block_idx,
+                    "delta": {"type": "text_delta", "text": text},
+                },
+            )
+
+        # Tool call deltas
+        tool_calls_delta = delta.get("tool_calls")
+        if isinstance(tool_calls_delta, list):
+            for tc in tool_calls_delta:
+                if not isinstance(tc, dict):
+                    continue
+                d_idx = int(tc.get("index") or 0)
+                if d_idx not in tool_block_indices:
+                    yield open_tool_block(d_idx, tc)
+                block_idx = tool_block_indices[d_idx]
+                fn = tc.get("function") or {}
+                args = fn.get("arguments")
+                if isinstance(args, str) and args:
+                    yield _sse_event(
+                        "content_block_delta",
+                        {
+                            "type": "content_block_delta",
+                            "index": block_idx,
+                            "delta": {
+                                "type": "input_json_delta",
+                                "partial_json": args,
+                            },
+                        },
+                    )
+
+        finish = choice.get("finish_reason")
+        if finish:
+            final_finish_reason = finish
+
+    # Close any open content blocks
+    if text_block_idx is not None:
+        yield close_block(text_block_idx)
+    for block_idx in tool_block_indices.values():
+        yield close_block(block_idx)
+
+    # message_delta with stop_reason and usage
+    stop_reason = _FINISH_TO_STOP.get(final_finish_reason or "", "end_turn")
+    yield _sse_event(
+        "message_delta",
+        {
+            "type": "message_delta",
+            "delta": {"stop_reason": stop_reason, "stop_sequence": None},
+            "usage": final_usage,
+        },
+    )
+    yield _sse_event("message_stop", {"type": "message_stop"})
+
+
+async def _post_and_stream(
+    base_url: str,
+    api_key: str,
+    payload: dict,
+    log_extra: dict[str, Any] | None,
+) -> tuple[httpx.AsyncClient, httpx.Response]:
+    """POST to upstream chat-completions and return (client, response) for
+    streaming. Caller is responsible for closing both."""
+    url = f"{base_url.rstrip('/')}/chat/completions"
+    client = httpx.AsyncClient(timeout=httpx.Timeout(120.0, read=120.0))
+    try:
+        request = client.build_request(
+            "POST",
+            url,
+            json=payload,
+            headers={
+                "Authorization": f"Bearer {api_key}",
+                "Content-Type": "application/json",
+                "Accept": "text/event-stream",
+            },
+        )
+        response = await client.send(request, stream=True)
+    except Exception as exc:
+        await client.aclose()
+        logger.error(
+            "Gemini messages dispatch HTTP error",
+            extra={"error": str(exc), "url": url, **(log_extra or {})},
+        )
+        raise UpstreamError(
+            f"Failed to reach Gemini upstream: {exc}", status_code=502
+        ) from exc
+
+    if response.status_code >= 400:
+        try:
+            body_bytes = await response.aread()
+        finally:
+            await response.aclose()
+            await client.aclose()
+        body_text = body_bytes.decode("utf-8", errors="replace")
+        logger.error(
+            "Gemini messages dispatch upstream error",
+            extra={
+                "status_code": response.status_code,
+                "body": body_text[:1000],
+                "url": url,
+                **(log_extra or {}),
+            },
+        )
+        raise UpstreamError(
+            f"Upstream error via gemini compat: {body_text}",
+            status_code=response.status_code,
+        )
+
+    return client, response
+
+
+async def dispatch_gemini_messages(
+    *,
+    request_body: bytes | None,
+    model_obj: Model,
+    base_url: str,
+    api_key: str,
+    transform_model_name: Callable[[str], str],
+    log_extra: dict[str, Any] | None = None,
+) -> tuple[bool, Any, str | None]:
+    """Dispatch a /v1/messages request to Gemini's OpenAI-compat endpoint
+    with thought-signature injection.
+
+    Returns ``(client_stream, result, requested_model)`` where ``result``
+    is either an ``AsyncIterator[bytes]`` of Anthropic-format SSE events
+    (for streaming clients) or an Anthropic Message dict (after the caller
+    aggregates).
+    """
+    if not request_body:
+        raise UpstreamError(
+            "Missing request body for /v1/messages", status_code=400
+        )
+
+    try:
+        body: dict = json.loads(request_body)
+    except json.JSONDecodeError as exc:
+        raise UpstreamError(
+            f"Invalid JSON in /v1/messages body: {exc}", status_code=400
+        ) from exc
+
+    body.pop("model", None)
+    client_stream = bool(body.pop("stream", False))
+
+    # Anthropic-Messages-only fields that don't translate to OpenAI
+    # Chat Completions. litellm's translator passes through unknown
+    # top-level fields verbatim and Gemini's compat surface 400s on
+    # unknown names like ``context_management`` / ``output_config``.
+    dropped: dict[str, Any] = {}
+    for field in ANTHROPIC_ONLY_FIELDS:
+        if field in body:
+            dropped[field] = body.pop(field)
+    if dropped:
+        logger.debug(
+            "Dropped anthropic-only fields before gemini compat dispatch",
+            extra={"dropped_keys": sorted(dropped.keys())},
+        )
+
+    requested_model = (
+        (model_obj.forwarded_model_id or model_obj.id) if model_obj else None
+    )
+    upstream_model = transform_model_name(model_obj.id)
+
+    openai_kwargs = _translate_anthropic_to_openai(body, upstream_model)
+
+    messages = openai_kwargs.get("messages") or []
+    if isinstance(messages, list):
+        inject_thought_signatures(messages)
+
+    # Disable Gemini's thinking pass; the dummy signature already lifts
+    # validation, but skipping thinking entirely avoids degraded model
+    # output and keeps tool-calling deterministic.
+    openai_kwargs.setdefault("reasoning_effort", "none")
+    openai_kwargs["stream"] = True
+    openai_kwargs["model"] = upstream_model
+    # OpenAI-compat backends (including Gemini's) only emit a final
+    # ``usage`` chunk when the request opts in via this flag. Without it
+    # the cost-calculation pipeline can't read real token counts and
+    # falls back to MaxCostData billing.
+    existing_stream_options = openai_kwargs.get("stream_options")
+    merged_stream_options = (
+        dict(existing_stream_options)
+        if isinstance(existing_stream_options, dict)
+        else {}
+    )
+    merged_stream_options.setdefault("include_usage", True)
+    openai_kwargs["stream_options"] = merged_stream_options
+
+    logger.info(
+        "Dispatching /v1/messages via gemini compat (httpx)",
+        extra={
+            "model": upstream_model,
+            "client_stream": client_stream,
+            "messages_with_tool_calls": sum(
+                1 for m in messages if isinstance(m, dict) and m.get("tool_calls")
+            ),
+            **(log_extra or {}),
+        },
+    )
+
+    http_client, response = await _post_and_stream(
+        base_url, api_key, openai_kwargs, log_extra
+    )
+
+    async def line_iter() -> AsyncGenerator[str, None]:
+        try:
+            async for line in response.aiter_lines():
+                yield line
+        finally:
+            await response.aclose()
+            await http_client.aclose()
+
+    anthropic_event_iter = _openai_chunks_to_anthropic_events(
+        line_iter(), requested_model
+    )
+
+    if not client_stream:
+        # Aggregate the Anthropic SSE byte stream into a single Message dict
+        # so the rest of the pipeline (cost calc, metadata injection,
+        # response building) can treat it identically to a non-streaming
+        # litellm response.
+        try:
+            aggregated = await aggregate_anthropic_events_to_message(
+                anthropic_event_iter
+            )
+        except Exception as exc:
+            logger.error(
+                "Failed to aggregate Gemini compat events into message",
+                extra={"error": str(exc), **(log_extra or {})},
+            )
+            raise UpstreamError(
+                f"Failed to aggregate upstream stream: {exc}",
+                status_code=502,
+            ) from exc
+        return client_stream, aggregated, requested_model
+
+    return client_stream, anthropic_event_iter, requested_model

routstr/upstream/generic.py

@@ -5,6 +5,12 @@ from typing import TYPE_CHECKING
 import httpx
 
 from .base import BaseUpstreamProvider
+from .pricing_resolver import (
+    FallbackPricingResolver,
+    ResolvedPricing,
+    _as_float,
+    estimate_context_length,
+)
 
 if TYPE_CHECKING:
     from ..core.db import UpstreamProviderRow
@@ -45,7 +51,7 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "GenericUpstreamProvider":
         return cls(
@@ -64,6 +70,40 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
             "platform_url": cls.platform_url,
         }
 
+    def _native_pricing(
+        self, model_id: str, model_spec: dict
+    ) -> ResolvedPricing | None:
+        """Read pricing/metadata from Venice's bespoke ``model_spec`` schema.
+
+        Returns ``None`` when the upstream reported no *usable* native price —
+        absent, non-numeric, negative, or both-zero — so the caller falls
+        through to the shared resolution chain instead of fabricating a number
+        or trusting a bogus one. This mirrors the money-safety guards the
+        litellm and OpenRouter rungs already apply: a both-zero price would
+        serve the model free, a negative one would credit the caller, and a
+        non-numeric string would otherwise throw and drop the whole catalog.
+        """
+        pricing_info = model_spec.get("pricing", {})
+        input_usd = _as_float(pricing_info.get("input", {}).get("usd"))
+        output_usd = _as_float(pricing_info.get("output", {}).get("usd"))
+        if input_usd is None or output_usd is None:
+            return None
+        if input_usd < 0 or output_usd < 0 or (input_usd == 0 and output_usd == 0):
+            return None
+
+        capabilities = model_spec.get("capabilities", {})
+        input_modalities = ["text"]
+        if capabilities.get("supportsVision", False):
+            input_modalities.append("image")
+
+        return ResolvedPricing(
+            prompt=input_usd / 1_000_000,
+            completion=output_usd / 1_000_000,
+            context_length=model_spec.get("availableContextTokens"),
+            source="native",
+            input_modalities=input_modalities,
+        )
+
     async def fetch_models(self) -> list[Model]:
         """Fetch models from upstream API using /models endpoint."""
         from ..payment.models import Architecture, Model, Pricing, TopProvider
@@ -78,6 +118,7 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
                 response.raise_for_status()
                 data = response.json()
 
+                resolver = FallbackPricingResolver()
                 models_list = []
                 for model_data in data.get("data", []):
                     model_id = model_data.get("id", "")
@@ -89,41 +130,44 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
                     owned_by = model_data.get("owned_by", "unknown")
                     model_spec = model_data.get("model_spec", {})
 
-                    context_length = 4096
-                    if model_spec.get("availableContextTokens"):
-                        context_length = model_spec["availableContextTokens"]
-                    elif any(
-                        pattern in model_id.lower() for pattern in ["32k", "32000"]
-                    ):
-                        context_length = 32768
-                    elif any(
-                        pattern in model_id.lower() for pattern in ["16k", "16000"]
-                    ):
-                        context_length = 16384
-                    elif any(pattern in model_id.lower() for pattern in ["8k", "8000"]):
-                        context_length = 8192
-                    elif "gpt-4" in model_id.lower():
-                        context_length = 8192
-                    elif "claude" in model_id.lower():
-                        context_length = 200000
+                    resolved = self._native_pricing(model_id, model_spec)
+                    if resolved is None:
+                        resolved = await resolver.resolve(model_id)
 
-                    pricing_info = model_spec.get("pricing", {})
-                    input_pricing = pricing_info.get("input", {})
-                    output_pricing = pricing_info.get("output", {})
+                    if resolved is None:
+                        # Fail closed: never invent a price. Import the model
+                        # disabled with a warning so the operator can price it
+                        # (the admin UI surfaces disabled remote models).
+                        logger.warning(
+                            f"No pricing source resolved for '{model_id}' from "
+                            f"{self.upstream_name}; importing it disabled",
+                            extra={"model_id": model_id, "base_url": self.base_url},
+                        )
+                        resolved = ResolvedPricing(
+                            prompt=0.0,
+                            completion=0.0,
+                            context_length=None,
+                            source="unresolved",
+                        )
+                        enabled = False
+                    else:
+                        enabled = True
 
-                    prompt_price = input_pricing.get("usd", 0.001) / 1000000
-                    completion_price = output_pricing.get("usd", 0.001) / 1000000
+                    # Prefer the source's own modality string (OpenRouter ships
+                    # one, e.g. "text+image->text"); otherwise derive it from the
+                    # captured input/output modalities in the same "in->out" shape
+                    # rather than flattening vision models to "text->text".
+                    modality = resolved.modality or (
+                        f"{'+'.join(resolved.input_modalities)}"
+                        f"->{'+'.join(resolved.output_modalities)}"
+                    )
 
-                    capabilities = model_spec.get("capabilities", {})
-                    input_modalities = ["text"]
-                    output_modalities = ["text"]
-
-                    if capabilities.get("supportsVision", False):
-                        input_modalities.append("image")
-
-                    modality = "text"
-                    if capabilities.get("supportsVision", False):
-                        modality = "text->text"
+                    # A source can carry a price but no context (e.g. a litellm
+                    # entry missing max_input_tokens); fall back to an id-based
+                    # estimate so we never persist a zero-length window.
+                    context_length = resolved.context_length or estimate_context_length(
+                        model_id
+                    )
 
                     spec_name = model_spec.get("name", model_name)
                     description = f"{spec_name}"
@@ -139,30 +183,33 @@ class GenericUpstreamProvider(BaseUpstreamProvider):
                             context_length=context_length,
                             architecture=Architecture(
                                 modality=modality,
-                                input_modalities=input_modalities,
-                                output_modalities=output_modalities,
-                                tokenizer="unknown",
-                                instruct_type=None,
+                                input_modalities=resolved.input_modalities,
+                                output_modalities=resolved.output_modalities,
+                                tokenizer=resolved.tokenizer,
+                                instruct_type=resolved.instruct_type,
                             ),
                             pricing=Pricing(
-                                prompt=prompt_price,
-                                completion=completion_price,
+                                prompt=resolved.prompt,
+                                completion=resolved.completion,
                                 request=0.0,
                                 image=0.0,
                                 web_search=0.0,
                                 internal_reasoning=0.0,
-                                max_prompt_cost=0.001,
-                                max_completion_cost=0.001,
-                                max_cost=0.001,
+                                input_cache_read=resolved.input_cache_read,
+                                input_cache_write=resolved.input_cache_write,
                             ),
                             sats_pricing=None,
                             per_request_limits=None,
                             top_provider=TopProvider(
                                 context_length=context_length,
-                                max_completion_tokens=context_length // 2,
-                                is_moderated=False,
+                                max_completion_tokens=(
+                                    resolved.max_completion_tokens
+                                    if resolved.max_completion_tokens is not None
+                                    else context_length // 2
+                                ),
+                                is_moderated=bool(resolved.is_moderated),
                             ),
-                            enabled=True,
+                            enabled=enabled,
                             upstream_provider_id=None,
                             canonical_slug=None,
                         )

routstr/upstream/groq.py

@@ -12,6 +12,7 @@ class GroqUpstreamProvider(BaseUpstreamProvider):
     provider_type = "groq"
     default_base_url = "https://api.groq.com/openai/v1"
     platform_url = "https://console.groq.com/keys"
+    litellm_provider_prefix = "groq/"
 
     def __init__(self, api_key: str, provider_fee: float = 1.01):
         super().__init__(
@@ -19,7 +20,7 @@ class GroqUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(cls, provider_row: "UpstreamProviderRow") -> "GroqUpstreamProvider":
+    def _build_from_row(cls, provider_row: "UpstreamProviderRow") -> "GroqUpstreamProvider":
         return cls(
             api_key=provider_row.api_key,
             provider_fee=provider_row.provider_fee,

routstr/upstream/helpers.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 import asyncio
 import os
 import re
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Callable
 
 if TYPE_CHECKING:
     from ..core.settings import Settings
@@ -12,6 +12,7 @@ from sqlmodel import select
 
 from ..core import get_logger
 from ..core.db import AsyncSession, ModelRow, UpstreamProviderRow, create_session
+from ..core.provider_slugs import allocate_unique_provider_slug
 from ..payment.models import Model
 from .base import BaseUpstreamProvider
 
@@ -93,12 +94,10 @@ async def get_all_models_with_overrides(
         provider_result = await session.exec(select(UpstreamProviderRow))
         providers_by_id = {p.id: p for p in provider_result.all()}
 
-        overrides_by_id: dict[str, tuple[ModelRow, float]] = {
-            row.id: (
+        overrides_by_key: dict[tuple[str, int], tuple[ModelRow, float]] = {
+            (row.id.lower(), row.upstream_provider_id): (
                 row,
-                providers_by_id[row.upstream_provider_id].provider_fee
-                if row.upstream_provider_id in providers_by_id
-                else 1.01,
+                providers_by_id[row.upstream_provider_id].provider_fee,
             )
             for row in override_rows
             if row.upstream_provider_id is not None
@@ -106,28 +105,43 @@ async def get_all_models_with_overrides(
             and providers_by_id[row.upstream_provider_id].enabled
         }
 
-    all_models: dict[str, Model] = {}
+    all_models: dict[tuple[str, str], Model] = {}
 
     for upstream in upstreams:
+        upstream_db_id = getattr(upstream, "db_id", None)
+        provider_key = (
+            f"db:{upstream_db_id}"
+            if isinstance(upstream_db_id, int)
+            else f"{getattr(upstream, 'provider_type', '')}|{getattr(upstream, 'base_url', '')}"
+        )
         for model in upstream.get_cached_models():
-            if model.id in overrides_by_id:
-                override_row, provider_fee = overrides_by_id[model.id]
-                all_models[model.id] = _row_to_model(
+            model_key = (
+                (model.id.lower(), upstream_db_id)
+                if isinstance(upstream_db_id, int)
+                else None
+            )
+            if model_key is not None and model_key in overrides_by_key:
+                override_row, provider_fee = overrides_by_key[model_key]
+                all_models[(model.id.lower(), provider_key)] = _row_to_model(
                     override_row, apply_provider_fee=True, provider_fee=provider_fee
                 )
             elif model.enabled:
-                all_models[model.id] = model
+                all_models[(model.id.lower(), provider_key)] = model
 
     return list(all_models.values())
 
 
 async def refresh_upstreams_models_periodically(
-    upstreams: list[BaseUpstreamProvider],
+    upstreams_provider: (
+        Callable[[], list[BaseUpstreamProvider]] | list[BaseUpstreamProvider]
+    ),
 ) -> None:
     """Background task to periodically refresh models cache for all providers.
 
     Args:
-        upstreams: List of upstream provider instances
+        upstreams_provider: Either a callable returning the live upstream list
+            (preferred — picks up providers added/changed via reinitialize_upstreams),
+            or a static list (legacy, will go stale after reinitialize_upstreams).
     """
     import asyncio
     import random
@@ -139,9 +153,14 @@ async def refresh_upstreams_models_periodically(
         logger.info("Provider models refresh disabled (interval <= 0)")
         return
 
+    def _resolve_upstreams() -> list[BaseUpstreamProvider]:
+        if callable(upstreams_provider):
+            return upstreams_provider()
+        return upstreams_provider
+
     while True:
         try:
-            for upstream in upstreams:
+            for upstream in _resolve_upstreams():
                 try:
                     await upstream.refresh_models_cache()
                 except Exception as e:
@@ -188,13 +207,14 @@ async def init_upstreams() -> list[BaseUpstreamProvider]:
         existing_providers = result.all()
 
         if not existing_providers:
-            logger.info(
-                "No upstream providers found in database, seeding from settings"
-            )
             await _seed_providers_from_settings(session, settings)
             await session.commit()
             result = await session.exec(select(UpstreamProviderRow))
             existing_providers = result.all()
+            if existing_providers:
+                logger.info(
+                    f"Seeded {len(existing_providers)} upstream providers from settings"
+                )
 
         async def _init_single_provider(
             provider_row: UpstreamProviderRow,
@@ -237,6 +257,7 @@ async def _seed_providers_from_settings(
 
     providers_to_add: list[UpstreamProviderRow] = []
     seeded_provider_keys: set[tuple[str, str]] = set()
+    reserved_slugs: set[str] = set()
 
     provider_classes_by_type = {
         cls.provider_type: cls
@@ -251,6 +272,7 @@ async def _seed_providers_from_settings(
         ("PERPLEXITY_API_KEY", "perplexity", None, None),
         ("FIREWORKS_API_KEY", "fireworks", None, None),
         ("XAI_API_KEY", "xai", None, None),
+        ("TINFOIL_API_KEY", "tinfoil", None, None),
     ]
 
     for env_key, provider_type, _, _ in env_mappings:
@@ -266,8 +288,13 @@ async def _seed_providers_from_settings(
                     )
                 )
                 if not result.first():
+                    slug = await allocate_unique_provider_slug(
+                        session, provider_type, reserved_slugs
+                    )
+                    reserved_slugs.add(slug)
                     providers_to_add.append(
                         UpstreamProviderRow(
+                            slug=slug,
                             provider_type=provider_type,
                             base_url=base_url,
                             api_key=api_key,
@@ -286,8 +313,13 @@ async def _seed_providers_from_settings(
             )
         )
         if not result.first():
+            slug = await allocate_unique_provider_slug(
+                session, "ollama", reserved_slugs
+            )
+            reserved_slugs.add(slug)
             providers_to_add.append(
                 UpstreamProviderRow(
+                    slug=slug,
                     provider_type="ollama",
                     base_url=ollama_base_url,
                     api_key=ollama_api_key,
@@ -307,8 +339,13 @@ async def _seed_providers_from_settings(
                 )
             )
             if not result.first():
+                slug = await allocate_unique_provider_slug(
+                    session, "azure", reserved_slugs
+                )
+                reserved_slugs.add(slug)
                 providers_to_add.append(
                     UpstreamProviderRow(
+                        slug=slug,
                         provider_type="azure",
                         base_url=base_url,
                         api_key=api_key,
@@ -329,8 +366,13 @@ async def _seed_providers_from_settings(
                 )
             )
             if not result.first():
+                slug = await allocate_unique_provider_slug(
+                    session, "custom", reserved_slugs
+                )
+                reserved_slugs.add(slug)
                 providers_to_add.append(
                     UpstreamProviderRow(
+                        slug=slug,
                         provider_type="custom",
                         base_url=base_url,
                         api_key=api_key,
@@ -343,7 +385,7 @@ async def _seed_providers_from_settings(
         session.add(provider)
         logger.info(
             f"Seeding {provider.provider_type} provider",  # type: ignore[str-format]
-            extra={"base_url": provider.base_url},
+            extra={"base_url": provider.base_url, "slug": provider.slug},
         )
 
 
@@ -378,9 +420,7 @@ def _instantiate_provider(
             return provider
 
         if provider_row.provider_type == "custom":
-            return BaseUpstreamProvider(
-                provider_row.base_url, provider_row.api_key, provider_row.provider_fee
-            )
+            return BaseUpstreamProvider.from_db_row(provider_row)
 
         logger.error(
             f"Unknown provider type: {provider_row.provider_type}",

routstr/upstream/litellm_routing.py

@@ -0,0 +1,162 @@
+"""Map an upstream `base_url` to the correct litellm provider prefix.
+
+Used by `BaseUpstreamProvider.get_litellm_provider_prefix` so that custom /
+generic provider rows (which inherit the base class default) get routed to
+the right litellm backend instead of falling back to `openai/`.
+
+The table is compiled from litellm 1.74's
+`litellm/litellm_core_utils/get_llm_provider_logic.py` (the
+`openai_compatible_endpoints` table) plus the providers documented at
+https://docs.litellm.ai/docs/providers. Substring match is used so that
+URLs with paths, ports, regional subdomains, etc. all resolve correctly.
+
+Order matters: more specific needles must appear before more generic ones
+(e.g. ``openai.azure.com`` before ``api.openai.com``).
+"""
+
+from __future__ import annotations
+
+import os
+from urllib.parse import urlsplit
+
+import litellm
+
+DEFAULT_PREFIX = "openai/"
+
+LITELLM_HOST_PREFIX_MAP: tuple[tuple[str, str], ...] = (
+    # Azure must win over api.openai.com because the host ends with
+    # `openai.azure.com` and we don't want it picked up as plain OpenAI.
+    ("openai.azure.com", "azure/"),
+    # Google
+    ("generativelanguage.googleapis.com", "gemini/"),
+    ("aiplatform.googleapis.com", "vertex_ai/"),
+    # First-class providers with native litellm prefixes
+    ("api.openai.com", "openai/"),
+    ("api.anthropic.com", "anthropic/"),
+    ("api.groq.com", "groq/"),
+    ("api.fireworks.ai", "fireworks_ai/"),
+    ("api.x.ai", "xai/"),
+    ("api.perplexity.ai", "perplexity/"),
+    ("openrouter.ai", "openrouter/"),
+    ("api.deepseek.com", "deepseek/"),
+    ("api.together.xyz", "together_ai/"),
+    ("codestral.mistral.ai", "codestral/"),
+    ("api.mistral.ai", "mistral/"),
+    ("api.cohere.com", "cohere_chat/"),
+    ("api.cohere.ai", "cohere_chat/"),
+    ("api.deepinfra.com", "deepinfra/"),
+    ("api.endpoints.anyscale.com", "anyscale/"),
+    ("api.cerebras.ai", "cerebras/"),
+    ("inference.baseten.co", "baseten/"),
+    ("api.sambanova.ai", "sambanova/"),
+    ("api.ai21.com", "ai21_chat/"),
+    ("api.friendli.ai", "friendliai/"),
+    ("api.galadriel.com", "galadriel/"),
+    ("api.llama.com", "meta_llama/"),
+    ("api.featherless.ai", "featherless_ai/"),
+    ("inference.api.nscale.com", "nscale/"),
+    ("dashscope-intl.aliyuncs.com", "dashscope/"),
+    ("api.moonshot.ai", "moonshot/"),
+    ("api.moonshot.cn", "moonshot/"),
+    ("api.minimax.io", "minimax/"),
+    ("api.minimaxi.com", "minimax/"),
+    ("platform.publicai.co", "publicai/"),
+    ("api.synthetic.new", "synthetic/"),
+    ("api.stima.tech", "apertis/"),
+    ("nano-gpt.com", "nano-gpt/"),
+    ("api.poe.com", "poe/"),
+    ("llm.chutes.ai", "chutes/"),
+    ("api.v0.dev", "v0/"),
+    ("api.lambda.ai", "lambda_ai/"),
+    ("api.hyperbolic.xyz", "hyperbolic/"),
+    ("ai-gateway.vercel.sh", "vercel_ai_gateway/"),
+    ("api.inference.wandb.ai", "wandb/"),
+    ("integrate.api.nvidia.com", "nvidia_nim/"),
+    ("api.studio.nebius.com", "nebius/"),
+    ("api.novita.ai", "novita/"),
+    ("ark.cn-beijing.volces.com", "volcengine/"),
+    ("api.voyageai.com", "voyage/"),
+    ("api.jina.ai", "jina_ai/"),
+    ("api.aimlapi.com", "aiml/"),
+    ("api.snowflakecomputing.com", "snowflake/"),
+    ("databricks.com", "databricks/"),
+    ("huggingface.co", "huggingface/"),
+)
+
+# Substrings that indicate an Ollama deployment regardless of port/scheme.
+OLLAMA_HOST_HINTS: tuple[str, ...] = (
+    "localhost:11434",
+    "127.0.0.1:11434",
+    "ollama",
+)
+
+
+def detect_litellm_prefix(
+    base_url: str | None, default: str = DEFAULT_PREFIX
+) -> str:
+    """Return the litellm provider prefix (`"<provider>/"`) for `base_url`.
+
+    Falls back to `default` when the host doesn't match any known provider.
+    The default is `openai/` because every unmatched OpenAI-compatible
+    server is, by definition, an OpenAI-compatible server.
+    """
+    if not base_url:
+        return default
+
+    parsed = urlsplit(base_url)
+    host = parsed.netloc.lower() or base_url.lower()
+
+    for needle, prefix in LITELLM_HOST_PREFIX_MAP:
+        if needle in host:
+            return prefix
+
+    if any(hint in host for hint in OLLAMA_HOST_HINTS):
+        return "ollama_chat/"
+
+    return default
+
+
+_configured = False
+
+
+def configure_litellm() -> None:
+    """Apply litellm global settings used by the messages-dispatch path.
+
+    Idempotent: safe to call from both app startup and module-level
+    initializers without side effects on the second invocation.
+
+    Settings applied:
+
+    * ``LITELLM_DEBUG=1`` enables litellm's verbose debug logger.
+    * Forces the Anthropic-messages adapter to call OpenAI Chat Completions
+      (POST ``/chat/completions``) instead of the Responses API (POST
+      ``/responses``) for ``openai/``-prefixed providers. OpenAI-compatible
+      upstreams like Google's generativelanguage compat endpoint expose
+      ``/chat/completions`` but not ``/responses``, which would 404. Set
+      ``LITELLM_USE_RESPONSES_API_FOR_ANTHROPIC_MESSAGES=1`` to opt out.
+    * Silently drops Anthropic-Messages-only parameters (``thinking``,
+      ``cache_control``, ``context_management``, ...) when translating to
+      providers that don't accept them, instead of raising
+      ``UnsupportedParamsError``. Set ``LITELLM_STRICT_PARAMS=1`` to opt
+      out.
+    """
+    global _configured
+    if _configured:
+        return
+
+    if os.getenv("LITELLM_DEBUG") == "1":
+        try:
+            litellm._turn_on_debug()  # type: ignore[no-untyped-call]
+        except Exception:
+            pass
+
+    if os.getenv("LITELLM_USE_RESPONSES_API_FOR_ANTHROPIC_MESSAGES") != "1":
+        try:
+            litellm.use_chat_completions_url_for_anthropic_messages = True
+        except Exception:
+            pass
+
+    if os.getenv("LITELLM_STRICT_PARAMS") != "1":
+        litellm.drop_params = True
+
+    _configured = True

routstr/upstream/messages_dispatch.py

@@ -0,0 +1,575 @@
+"""Pure helpers for translating ``/v1/messages`` to upstream chat completions
+via litellm.
+
+This module owns the litellm/Anthropic-Messages translation layer:
+
+* SSE parsing (``parse_sse_blocks``, ``events_from_chunk``)
+* Payload coercion (``coerce_litellm_payload``)
+* Stream aggregation (``aggregate_anthropic_events_to_message``) — drains
+  a streamed Anthropic event sequence into a single Message dict
+* Per-event annotation for streaming (``annotate_event``,
+  ``stream_annotated_events``) — handles the model-rewrite + token-tally
+  bookkeeping shared by the bearer-key and x-cashu streaming paths
+* The dispatch entry point (``dispatch_anthropic_messages``)
+* Refund math (``compute_refund``)
+
+Nothing in here touches ``BaseUpstreamProvider``; the thin instance methods
+on the provider class forward to these functions and only retain logic that
+genuinely needs ``self`` (cost adjustment, metadata injection, refund
+sending).
+"""
+
+from __future__ import annotations
+
+import json
+from collections.abc import AsyncGenerator, AsyncIterator
+from typing import Any, Callable, NamedTuple, cast
+
+import litellm
+
+from ..core import get_logger
+from ..core.exceptions import UpstreamError
+from ..core.redaction import redact_org_ids
+from ..payment.models import Model
+from .rate_limit import classify_rate_limit
+
+logger = get_logger(__name__)
+
+# Anthropic-Messages-only fields that don't translate to OpenAI
+# Chat Completions. ``litellm.drop_params`` only filters *known*
+# unsupported params; these newer/extension fields get passed through
+# verbatim and the upstream rejects them with a 400. Pop them here so the
+# request reaches the upstream cleanly.
+ANTHROPIC_ONLY_FIELDS: tuple[str, ...] = (
+    "thinking",
+    "cache_control",
+    "context_management",
+    "output_config",
+    "mcp_servers",
+    "service_tier",
+    "anthropic_version",
+    "anthropic_beta",
+)
+
+
+def coerce_litellm_payload(payload: object) -> dict:
+    """Convert a litellm event into a plain dict.
+
+    Non-streaming responses come back as Anthropic-shaped pydantic models
+    or dicts. Streaming may yield raw bytes/str (SSE-encoded); those go
+    through ``events_from_chunk`` instead, not here.
+    """
+    if isinstance(payload, dict):
+        return dict(payload)
+    if hasattr(payload, "model_dump"):
+        return cast(dict, payload.model_dump())
+    raise TypeError(f"Cannot coerce {type(payload).__name__} to dict")
+
+
+def parse_sse_blocks(buffer: bytes) -> tuple[list[dict], bytes]:
+    """Parse complete SSE event blocks out of a byte buffer.
+
+    Returns (events, remaining_buffer). Events are JSON objects parsed from
+    one or more ``data:`` lines per block. Comments, blank lines, and
+    ``[DONE]`` sentinels are ignored. A trailing partial block is preserved
+    in remaining_buffer.
+    """
+    events: list[dict] = []
+    while True:
+        sep = buffer.find(b"\n\n")
+        if sep < 0:
+            sep_rn = buffer.find(b"\r\n\r\n")
+            if sep_rn < 0:
+                break
+            block = buffer[:sep_rn]
+            buffer = buffer[sep_rn + 4 :]
+        else:
+            block = buffer[:sep]
+            buffer = buffer[sep + 2 :]
+
+        data_lines: list[str] = []
+        for raw_line in block.replace(b"\r\n", b"\n").split(b"\n"):
+            line = raw_line.decode("utf-8", errors="replace")
+            if line.startswith(":"):
+                continue
+            if line.startswith("data:"):
+                data_lines.append(line[5:].lstrip())
+        if not data_lines:
+            continue
+        payload = "\n".join(data_lines).strip()
+        if not payload or payload == "[DONE]":
+            continue
+        try:
+            obj = json.loads(payload)
+        except json.JSONDecodeError:
+            continue
+        if isinstance(obj, dict):
+            events.append(obj)
+    return events, buffer
+
+
+def events_from_chunk(
+    chunk: object, sse_buffer: bytes
+) -> tuple[list[dict], bytes]:
+    """Normalize a stream chunk into one or more event dicts.
+
+    ``litellm.anthropic.messages.acreate(stream=True)`` yields raw SSE
+    bytes in practice; some adapters yield strings or typed events. Handle
+    all three.
+    """
+    if isinstance(chunk, (bytes, bytearray)):
+        sse_buffer += bytes(chunk)
+        events, sse_buffer = parse_sse_blocks(sse_buffer)
+        return events, sse_buffer
+    if isinstance(chunk, str):
+        sse_buffer += chunk.encode("utf-8")
+        events, sse_buffer = parse_sse_blocks(sse_buffer)
+        return events, sse_buffer
+    return [coerce_litellm_payload(chunk)], sse_buffer
+
+
+async def aggregate_anthropic_events_to_message(
+    iterator: AsyncIterator[Any],
+) -> dict:
+    """Drain an Anthropic-Messages event iterator into a single Message dict.
+
+    Produces the shape ``litellm.anthropic.messages.acreate(stream=False)``
+    would have returned. Used to transparently stream from upstream while
+    still returning a non-streaming response to the client. Lets us
+    sidestep upstream quirks (e.g. Fireworks rejects ``max_tokens > 4096``
+    unless ``stream=true``) without leaking that into client-visible
+    behavior.
+    """
+    sse_buffer = b""
+    message: dict = {}
+    blocks: list[dict] = []
+    partial_json: dict[int, str] = {}
+    final_stop_reason: str | None = None
+    final_stop_sequence: str | None = None
+    final_usage: dict[str, Any] = {}
+    final_model: str | None = None
+
+    async for chunk in iterator:
+        events, sse_buffer = events_from_chunk(chunk, sse_buffer)
+        for event in events:
+            etype = event.get("type")
+            if etype == "message_start":
+                raw = event.get("message") or {}
+                if isinstance(raw, dict):
+                    message = dict(raw)
+                    existing = message.get("content")
+                    blocks = list(existing) if isinstance(existing, list) else []
+                    usage = message.get("usage")
+                    if isinstance(usage, dict):
+                        final_usage = dict(usage)
+                    if isinstance(message.get("model"), str):
+                        final_model = message["model"]
+            elif etype == "content_block_start":
+                idx = int(event.get("index") or 0)
+                cb = event.get("content_block") or {}
+                cb_dict = dict(cb) if isinstance(cb, dict) else {}
+                while len(blocks) <= idx:
+                    blocks.append({})
+                blocks[idx] = cb_dict
+            elif etype == "content_block_delta":
+                idx = int(event.get("index") or 0)
+                if idx >= len(blocks):
+                    continue
+                delta = event.get("delta") or {}
+                if not isinstance(delta, dict):
+                    continue
+                dtype = delta.get("type")
+                block = blocks[idx]
+                if dtype == "text_delta":
+                    block["text"] = (block.get("text") or "") + (
+                        delta.get("text") or ""
+                    )
+                elif dtype == "input_json_delta":
+                    partial_json[idx] = partial_json.get(idx, "") + (
+                        delta.get("partial_json") or ""
+                    )
+                elif dtype == "thinking_delta":
+                    block["thinking"] = (block.get("thinking") or "") + (
+                        delta.get("thinking") or ""
+                    )
+                elif dtype == "signature_delta":
+                    block["signature"] = (block.get("signature") or "") + (
+                        delta.get("signature") or ""
+                    )
+            elif etype == "content_block_stop":
+                idx = int(event.get("index") or 0)
+                raw_json = partial_json.pop(idx, None)
+                if raw_json is not None and idx < len(blocks):
+                    try:
+                        blocks[idx]["input"] = (
+                            json.loads(raw_json) if raw_json else {}
+                        )
+                    except json.JSONDecodeError:
+                        blocks[idx]["input"] = raw_json
+            elif etype == "message_delta":
+                delta = event.get("delta") or {}
+                if isinstance(delta, dict):
+                    if "stop_reason" in delta:
+                        final_stop_reason = delta.get("stop_reason")
+                    if "stop_sequence" in delta:
+                        final_stop_sequence = delta.get("stop_sequence")
+                usage = event.get("usage")
+                if isinstance(usage, dict):
+                    final_usage.update(usage)
+            # message_stop: nothing to merge
+
+    if not message:
+        # Upstream returned no message_start; expose what we can so the
+        # client at least sees the assembled content.
+        message = {
+            "id": "",
+            "type": "message",
+            "role": "assistant",
+            "content": [],
+        }
+
+    message["content"] = blocks
+    if final_model and not message.get("model"):
+        message["model"] = final_model
+    if final_stop_reason is not None:
+        message["stop_reason"] = final_stop_reason
+    if final_stop_sequence is not None:
+        message["stop_sequence"] = final_stop_sequence
+    if final_usage:
+        existing_usage = message.get("usage")
+        merged = dict(existing_usage) if isinstance(existing_usage, dict) else {}
+        merged.update(final_usage)
+        message["usage"] = merged
+    return message
+
+
+class AnnotatedEvent(NamedTuple):
+    """One Anthropic SSE event after model-rewrite + token-tally bookkeeping.
+
+    ``sse_bytes`` is the wire-ready ``event:`` / ``data:`` block; the two
+    streaming paths in ``BaseUpstreamProvider`` consume ``sse_bytes`` plus
+    the tallies and only differ in whether they stream live or buffer
+    first.
+
+    ``cache_read_input_tokens`` and ``cache_creation_input_tokens`` are
+    surfaced separately so the cost path can price them against the cache
+    rate rather than fold them silently into the regular input bucket.
+
+    ``total_cost`` / ``input_cost`` / ``output_cost`` carry any
+    USD cost figures the upstream attached to this event (from
+    ``usage.cost``, ``usage.total_cost``, or ``usage.cost_details``) so the
+    streaming paths can re-embed them in the rebuilt ``usage`` dict and let
+    ``calculate_cost`` convert directly USD→sats instead of falling back to
+    token-based math.
+    """
+
+    event: dict
+    sse_bytes: bytes
+    input_tokens: int
+    output_tokens: int
+    cache_read_input_tokens: int
+    cache_creation_input_tokens: int
+    total_cost: float
+    input_cost: float
+    output_cost: float
+    model: str | None
+
+
+def annotate_event(event: dict, requested_model: str | None) -> AnnotatedEvent:
+    """Rewrite ``model`` fields and extract per-event token / model info.
+
+    Mutates ``event`` in place when ``requested_model`` is set so the
+    upstream's true model name doesn't leak to the client.
+    """
+    if requested_model:
+        msg = event.get("message")
+        if isinstance(msg, dict) and "model" in msg:
+            msg["model"] = requested_model
+        if "model" in event:
+            event["model"] = requested_model
+
+    in_tokens = 0
+    out_tokens = 0
+    cache_read_tokens = 0
+    cache_create_tokens = 0
+    total_cost = 0.0
+    input_cost = 0.0
+    output_cost = 0.0
+    model: str | None = None
+
+    def _coerce_float(value: object) -> float:
+        if value is None or isinstance(value, bool):
+            return 0.0
+        if not isinstance(value, (int, float, str)):
+            return 0.0
+        try:
+            return max(0.0, float(value))
+        except (TypeError, ValueError):
+            return 0.0
+
+    def _accumulate(usage: dict) -> None:
+        nonlocal in_tokens, out_tokens, cache_read_tokens, cache_create_tokens
+        nonlocal total_cost, input_cost, output_cost
+        in_tokens += int(usage.get("input_tokens") or 0)
+        out_tokens += int(usage.get("output_tokens") or 0)
+        cache_read_tokens += int(usage.get("cache_read_input_tokens") or 0)
+        cache_create_tokens += int(usage.get("cache_creation_input_tokens") or 0)
+        total_cost += _coerce_float(usage.get("total_cost"))
+        input_cost += _coerce_float(usage.get("input_cost"))
+        output_cost += _coerce_float(usage.get("output_cost"))
+
+    msg_for_meta = event.get("message")
+    if isinstance(msg_for_meta, dict):
+        if msg_for_meta.get("model"):
+            model = str(msg_for_meta["model"])
+        usage = msg_for_meta.get("usage")
+        if isinstance(usage, dict):
+            _accumulate(usage)
+
+    if isinstance(event.get("usage"), dict):
+        _accumulate(event["usage"])
+
+    # Some upstreams (notably OpenRouter-style proxies) attach cost fields
+    # directly at the event root rather than inside ``usage``.
+    for field in ("total_cost", "cost"):
+        total_cost = max(total_cost, _coerce_float(event.get(field)))
+    input_cost = max(input_cost, _coerce_float(event.get("input_cost")))
+    output_cost = max(output_cost, _coerce_float(event.get("output_cost")))
+    root_cost_details = event.get("cost_details")
+    if isinstance(root_cost_details, dict):
+        total_cost = max(
+            total_cost,
+            _coerce_float(root_cost_details.get("total_cost")),
+        )
+        input_cost = max(
+            input_cost,
+            _coerce_float(root_cost_details.get("input_cost")),
+        )
+        output_cost = max(
+            output_cost,
+            _coerce_float(root_cost_details.get("output_cost")),
+        )
+
+    event_type = str(event.get("type") or "")
+    payload = json.dumps(event)
+    if event_type:
+        sse_bytes = f"event: {event_type}\ndata: {payload}\n\n".encode()
+    else:
+        sse_bytes = f"data: {payload}\n\n".encode()
+
+    return AnnotatedEvent(
+        event,
+        sse_bytes,
+        in_tokens,
+        out_tokens,
+        cache_read_tokens,
+        cache_create_tokens,
+        total_cost,
+        input_cost,
+        output_cost,
+        model,
+    )
+
+
+async def stream_annotated_events(
+    iterator: AsyncIterator[Any],
+    requested_model: str | None,
+) -> AsyncGenerator[AnnotatedEvent, None]:
+    """Yield annotated, SSE-serialized events from a litellm stream.
+
+    Both streaming paths in ``BaseUpstreamProvider`` consume this; the only
+    divergence between them — yield-as-you-go vs buffer-then-replay — stays
+    in the caller.
+    """
+    sse_buffer = b""
+    async for chunk in iterator:
+        events, sse_buffer = events_from_chunk(chunk, sse_buffer)
+        for event in events:
+            yield annotate_event(event, requested_model)
+
+
+def embed_usd_costs(
+    usage: dict,
+    total_cost: float,
+    input_cost: float,
+    output_cost: float,
+) -> None:
+    """Mutate ``usage`` so ``calculate_cost`` will pick up the USD totals.
+
+    Mirrors the upstream shape: when any USD figure is present, attach
+    ``cost`` (used by the simple-fallback branch in ``calculate_cost``) and
+    a ``cost_details`` block (used by the preferred branch — also gives the
+    input/output USD split when we have one).
+    """
+    if total_cost <= 0 and input_cost <= 0 and output_cost <= 0:
+        return
+
+    cost_details: dict[str, float] = {}
+    effective_total = total_cost
+    if effective_total <= 0 and (input_cost > 0 or output_cost > 0):
+        effective_total = input_cost + output_cost
+
+    if effective_total > 0:
+        cost_details["total_cost"] = effective_total
+        usage["cost"] = effective_total
+    if input_cost > 0:
+        cost_details["input_cost"] = input_cost
+    if output_cost > 0:
+        cost_details["output_cost"] = output_cost
+    if cost_details:
+        usage["cost_details"] = cost_details
+
+
+def compute_refund(amount: int, unit: str, cost_msats: int) -> int:
+    if unit == "msat":
+        return amount - cost_msats
+    if unit == "sat":
+        return amount - (cost_msats + 999) // 1000
+    raise ValueError(f"Invalid unit: {unit}")
+
+
+async def dispatch_anthropic_messages(
+    *,
+    request_body: bytes | None,
+    model_obj: Model,
+    base_url: str,
+    api_key: str,
+    provider_prefix: str,
+    transform_model_name: Callable[[str], str],
+    log_extra: dict[str, Any] | None = None,
+) -> tuple[bool, Any, str | None]:
+    """Call ``litellm.anthropic.messages.acreate`` and return
+    ``(client_stream, result, requested_model)``.
+
+    Shared by the bearer-key and x-cashu paths. Raises :class:`UpstreamError`
+    on bad input or upstream failure.
+    """
+    if not request_body:
+        raise UpstreamError(
+            "Missing request body for /v1/messages", status_code=400
+        )
+
+    try:
+        body: dict = json.loads(request_body)
+    except json.JSONDecodeError as exc:
+        raise UpstreamError(
+            f"Invalid JSON in /v1/messages body: {exc}", status_code=400
+        ) from exc
+
+    body.pop("model", None)
+    # `stream` here is what the **client** asked for. Upstream is always
+    # streamed (see `upstream_stream` below); when the client asked for a
+    # non-streaming response we drain and aggregate the events into a
+    # single Anthropic Message dict before returning. This sidesteps
+    # provider-specific non-streaming caps (e.g. Fireworks rejects
+    # `max_tokens > 4096` unless `stream=true`).
+    client_stream = bool(body.pop("stream", False))
+    upstream_stream = True
+
+    dropped: dict[str, Any] = {}
+    for field in ANTHROPIC_ONLY_FIELDS:
+        if field in body:
+            dropped[field] = body.pop(field)
+    if dropped:
+        logger.debug(
+            "Dropped anthropic-only fields before litellm dispatch",
+            extra={"dropped_keys": sorted(dropped.keys())},
+        )
+
+    # Convention: `model.id` is the canonical upstream model name;
+    # `forwarded_model_id` is the public alias the internal API exposes
+    # and echoes back to the client.
+    requested_model = (
+        (model_obj.forwarded_model_id or model_obj.id) if model_obj else None
+    )
+    upstream_model = transform_model_name(model_obj.id)
+    litellm_model = f"{provider_prefix}{upstream_model}"
+
+    kwargs: dict = {
+        "model": litellm_model,
+        "api_base": base_url,
+        "api_key": api_key,
+        "stream": upstream_stream,
+        **body,
+    }
+
+    logger.info(
+        "Dispatching /v1/messages via litellm",
+        extra={
+            "model": litellm_model,
+            "resolved_provider": provider_prefix.rstrip("/"),
+            "client_stream": client_stream,
+            "upstream_stream": upstream_stream,
+            **(log_extra or {}),
+        },
+    )
+
+    try:
+        result = await litellm.anthropic.messages.acreate(**kwargs)
+    except Exception as exc:
+        raw_message = getattr(exc, "message", None) or str(exc) or repr(exc)
+        # Redact provider account identifiers before the message reaches logs
+        # or the surfaced error.
+        exc_message = redact_org_ids(raw_message)
+        exc_status = getattr(exc, "status_code", None)
+        exc_response = getattr(exc, "response", None)
+        response_text = None
+        if exc_response is not None:
+            try:
+                response_text = redact_org_ids(
+                    getattr(exc_response, "text", str(exc_response))
+                )
+            except Exception:
+                response_text = "<unreadable>"
+        status_for_classify = exc_status if isinstance(exc_status, int) else 502
+        rate_limit = classify_rate_limit(
+            status_for_classify, exc_message, getattr(exc, "headers", None)
+        )
+        logger.error(
+            "litellm dispatch failed",
+            extra={
+                "error": exc_message,
+                "error_type": type(exc).__name__,
+                "status_code": exc_status,
+                "error_code": rate_limit.code if rate_limit else None,
+                "llm_provider": getattr(exc, "llm_provider", None),
+                "body": redact_org_ids(str(getattr(exc, "body", "") or "")) or None,
+                "response_text": response_text,
+                "model": litellm_model,
+                "api_base": base_url,
+            },
+        )
+        raise UpstreamError(
+            f"Upstream error via litellm: {exc_message}",
+            status_code=status_for_classify,
+            code=rate_limit.code if rate_limit else None,
+            details=rate_limit.as_details() if rate_limit else None,
+        ) from exc
+
+    if not client_stream and hasattr(result, "__aiter__"):
+        # Client asked for a non-streaming response but we always stream
+        # from upstream — drain the events into a single Anthropic Message
+        # dict so the rest of the pipeline can treat it as if upstream had
+        # returned non-streaming. Some litellm adapters return a
+        # non-streaming dict even when ``stream=True``; in that case,
+        # leave the result as-is.
+        try:
+            aggregated: Any = await aggregate_anthropic_events_to_message(
+                cast(AsyncIterator[Any], result)
+            )
+        except Exception as exc:
+            logger.error(
+                "Failed to aggregate streamed events into message",
+                extra={
+                    "error": str(exc),
+                    "error_type": type(exc).__name__,
+                    "model": litellm_model,
+                },
+            )
+            raise UpstreamError(
+                f"Failed to aggregate upstream stream: {exc}",
+                status_code=502,
+            ) from exc
+        return client_stream, aggregated, requested_model
+
+    return client_stream, result, requested_model

routstr/upstream/ollama.py

@@ -3,13 +3,11 @@ from __future__ import annotations
 from typing import TYPE_CHECKING
 
 import httpx
-from fastapi import Request
-from fastapi.responses import Response, StreamingResponse
 
 from .base import BaseUpstreamProvider
 
 if TYPE_CHECKING:
-    from ..core.db import ApiKey, AsyncSession, UpstreamProviderRow
+    from ..core.db import UpstreamProviderRow
     from ..payment.models import Model
 
 from ..core.logging import get_logger
@@ -23,6 +21,7 @@ class OllamaUpstreamProvider(BaseUpstreamProvider):
     provider_type = "ollama"
     default_base_url = "http://localhost:11434"
     platform_url = None
+    litellm_provider_prefix = "ollama_chat/"
 
     def __init__(
         self,
@@ -44,7 +43,7 @@ class OllamaUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "OllamaUpstreamProvider":
         return cls(
@@ -67,38 +66,11 @@ class OllamaUpstreamProvider(BaseUpstreamProvider):
         """Strip 'ollama/' prefix for Ollama API compatibility."""
         return model_id.removeprefix("ollama/")
 
-    async def forward_request(
-        self,
-        request: Request,
-        path: str,
-        headers: dict,
-        request_body: bytes | None,
-        key: ApiKey,
-        max_cost_for_model: int,
-        session: AsyncSession,
-        model_obj: Model,
-    ) -> Response | StreamingResponse:
-        """Override to use OpenAI-compatible endpoint for proxy requests."""
-        if path.startswith("v1/"):
-            path = path.replace("v1/", "")
-
-        original_base_url = self.base_url
-        self.base_url = f"{self.base_url}/v1"
-
-        try:
-            result = await super().forward_request(
-                request,
-                path,
-                headers,
-                request_body,
-                key,
-                max_cost_for_model,
-                session,
-                model_obj,
-            )
-            return result
-        finally:
-            self.base_url = original_base_url
+    def get_request_base_url(
+        self, path: str, model_obj: Model | None = None
+    ) -> str:
+        """Route proxy traffic through Ollama's OpenAI-compatible /v1 endpoint."""
+        return f"{self.base_url.rstrip('/')}/v1"
 
     async def fetch_models(self) -> list[Model]:
         """Fetch models from Ollama API using /api/tags endpoint."""
@@ -213,7 +185,7 @@ class OllamaUpstreamProvider(BaseUpstreamProvider):
             except Exception:
                 self._models_cache = models_with_fees
 
-            self._models_by_id = {m.id: m for m in self._models_cache}
+            self._models_by_id = {m.forwarded_model_id or m.id: m for m in self._models_cache}
             logger.info(
                 f"Refreshed models cache for {self.base_url}",
                 extra={"model_count": len(models)},

routstr/upstream/openai.py

@@ -20,7 +20,7 @@ class OpenAIUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "OpenAIUpstreamProvider":
         return cls(

routstr/upstream/openrouter.py

@@ -15,6 +15,35 @@ class OpenRouterUpstreamProvider(BaseUpstreamProvider):
     provider_type = "openrouter"
     default_base_url = "https://openrouter.ai/api/v1"
     platform_url = "https://openrouter.ai/settings/keys"
+    supports_anthropic_messages = True
+    litellm_provider_prefix = "openrouter/"
+
+    def _apply_provider_field(self, response_json: object) -> None:
+        """Stamp the ``provider`` field for OpenRouter responses.
+
+        OpenRouter is a router, not the real serving provider, so a bare
+        ``"openrouter"`` value carries no useful information. Rules:
+
+        - Real upstream sub-provider (e.g. ``"GMICloud"``) -> ``"openrouter:GMICloud"``.
+        - Missing sub-provider, or one that merely echoes ``"openrouter"`` ->
+          ``"unknown"``.
+        - Idempotent: re-stamping never produces ``"openrouter:openrouter:..."``;
+          the ``openrouter:`` prefix appears at most once.
+        """
+        if not isinstance(response_json, dict):
+            return
+        provider_type = (self.provider_type or "").strip()
+        existing = response_json.get("provider")
+        sub = existing.strip() if isinstance(existing, str) else ""
+        # Strip any already-applied "openrouter:" prefixes (idempotency).
+        prefix = f"{provider_type}:"
+        while sub.lower().startswith(prefix.lower()):
+            sub = sub[len(prefix) :].strip()
+        # No real sub-provider, or it just echoes our own router name.
+        if not sub or sub.lower() == provider_type.lower():
+            response_json["provider"] = "unknown"
+            return
+        response_json["provider"] = f"{provider_type}:{sub}"
 
     def __init__(self, api_key: str, provider_fee: float = 1.06):
         """Initialize OpenRouter provider with API key.
@@ -28,7 +57,7 @@ class OpenRouterUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "OpenRouterUpstreamProvider":
         return cls(

routstr/upstream/perplexity.py

@@ -13,6 +13,7 @@ class PerplexityUpstreamProvider(BaseUpstreamProvider):
     provider_type = "perplexity"
     default_base_url = "https://api.perplexity.ai/"
     platform_url = "https://www.perplexity.ai/account/api/keys"
+    litellm_provider_prefix = "perplexity/"
 
     def __init__(self, api_key: str, provider_fee: float = 1.01):
         super().__init__(
@@ -22,7 +23,7 @@ class PerplexityUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "PerplexityUpstreamProvider":
         return cls(

routstr/upstream/ppqai.py

@@ -1,13 +1,14 @@
 from __future__ import annotations
 
-from typing import TYPE_CHECKING
+from typing import TYPE_CHECKING, Optional
 
 import httpx
-from pydantic import BaseModel
+from pydantic.v1 import BaseModel, Field
 
 from ..core.logging import get_logger
 from ..payment.models import Architecture, Model, Pricing, async_fetch_openrouter_models
 from .base import BaseUpstreamProvider, TopupData
+from .ehbp import EHBPForwardingTarget
 
 if TYPE_CHECKING:
     from ..core.db import UpstreamProviderRow
@@ -16,18 +17,20 @@ logger = get_logger(__name__)
 
 
 class PPQAIModelPricing(BaseModel):
-    ui: dict[str, float]
-    api: dict[str, float]
+    ui: Optional[dict[str, float]] = None
+    api: Optional[dict[str, float]] = None
+    input_per_1M_tokens: Optional[float] = Field(None, alias="input_per_1M_tokens")
+    output_per_1M_tokens: Optional[float] = Field(None, alias="output_per_1M_tokens")
 
 
 class PPQAIModel(BaseModel):
     id: str
-    provider: str
+    provider: Optional[str] = None
     name: str
     created_at: int
     context_length: int
     pricing: PPQAIModelPricing
-    popular: bool
+    popular: bool = False
 
 
 class PPQAIUpstreamProvider(BaseUpstreamProvider):
@@ -37,6 +40,10 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
     default_base_url = "https://api.ppq.ai"
     platform_url = "https://ppq.ai/api-docs"
     IGNORED_MODEL_IDS: list[str] = ["auto"]
+    # PPQ.AI has a private encrypted endpoint, but this proxy currently has no
+    # provider-attested usage extractor/model binding for it. Keep EHBP disabled
+    # until a ConfidentialInferenceProfile can bill it without max-cost fallback.
+    supports_ehbp = False
 
     def __init__(self, api_key: str, provider_fee: float = 1.0):
         super().__init__(
@@ -44,7 +51,7 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(
+    def _build_from_row(
         cls, provider_row: "UpstreamProviderRow"
     ) -> "PPQAIUpstreamProvider":
         return cls(
@@ -68,6 +75,20 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
     def transform_model_name(self, model_id: str) -> str:
         return model_id
 
+    def get_ehbp_forwarding_target(
+        self, path: str, model_obj: Model
+    ) -> EHBPForwardingTarget:
+        """Return the PPQ.AI private enclave target for EHBP requests.
+
+        PPQ.AI exposes EHBP-aware inference under /private/v1/... separate
+        from the public /v1/... endpoint. The encrypted body remains opaque to
+        Routstr, so PPQ.AI also needs X-Private-Model for routing/billing.
+        """
+        return EHBPForwardingTarget(
+            url=f"{self.base_url.rstrip('/')}/private/{path.lstrip('/')}",
+            headers={"X-Private-Model": model_obj.forwarded_model_id or model_obj.id},
+        )
+
     @classmethod
     async def create_account_static(cls) -> dict[str, object]:
         """Create a new PPQ.AI account without requiring an instance.
@@ -134,31 +155,54 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
                         )
 
                         if or_model:
-                            if input_price := ppqai_model.pricing.api.get(
-                                "input_per_1M"
-                            ):
+                            input_price = None
+                            if ppqai_model.pricing.api:
+                                input_price = ppqai_model.pricing.api.get(
+                                    "input_per_1M"
+                                )
+                            elif ppqai_model.pricing.input_per_1M_tokens:
+                                input_price = ppqai_model.pricing.input_per_1M_tokens
+
+                            if input_price is not None:
                                 or_model.pricing.prompt = input_price / 1_000_000
-                            if output_price := ppqai_model.pricing.api.get(
-                                "output_per_1M"
-                            ):
+
+                            output_price = None
+                            if ppqai_model.pricing.api:
+                                output_price = ppqai_model.pricing.api.get(
+                                    "output_per_1M"
+                                )
+                            elif ppqai_model.pricing.output_per_1M_tokens:
+                                output_price = ppqai_model.pricing.output_per_1M_tokens
+
+                            if output_price is not None:
                                 or_model.pricing.completion = output_price / 1_000_000
+
                             if cl := ppqai_model.context_length:
                                 or_model.context_length = cl
                             models.append(or_model)
                         else:
-                            input_price = ppqai_model.pricing.api.get(
-                                "input_per_1M", 0.0
-                            )
-                            output_price = ppqai_model.pricing.api.get(
-                                "output_per_1M", 0.0
-                            )
+                            input_price = 0.0
+                            if ppqai_model.pricing.api:
+                                input_price = ppqai_model.pricing.api.get(
+                                    "input_per_1M", 0.0
+                                )
+                            elif ppqai_model.pricing.input_per_1M_tokens:
+                                input_price = ppqai_model.pricing.input_per_1M_tokens
+
+                            output_price = 0.0
+                            if ppqai_model.pricing.api:
+                                output_price = ppqai_model.pricing.api.get(
+                                    "output_per_1M", 0.0
+                                )
+                            elif ppqai_model.pricing.output_per_1M_tokens:
+                                output_price = ppqai_model.pricing.output_per_1M_tokens
 
                             models.append(
                                 Model(
                                     id=ppqai_model.id,
                                     name=ppqai_model.name,
                                     created=ppqai_model.created_at // 1000,
-                                    description=f"{ppqai_model.provider} model",
+                                    description=f"{ppqai_model.provider or 'PPQ.AI'} model",
                                     context_length=ppqai_model.context_length,
                                     architecture=Architecture(
                                         modality="text->text",
@@ -204,17 +248,14 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
                 f"Disabling PPQ.AI provider ({self.base_url}) due to insufficient balance",
                 extra={"error": error_message},
             )
-            from sqlmodel import select
-
             from ..core.db import UpstreamProviderRow, create_session
 
             async with create_session() as session:
-                statement = select(UpstreamProviderRow).where(
-                    UpstreamProviderRow.base_url == self.base_url,
-                    UpstreamProviderRow.api_key == self.api_key,
+                provider = (
+                    await session.get(UpstreamProviderRow, self.db_id)
+                    if self.db_id is not None
+                    else None
                 )
-                result = await session.exec(statement)
-                provider = result.first()
 
                 if provider:
                     provider.enabled = False
@@ -283,7 +324,6 @@ class PPQAIUpstreamProvider(BaseUpstreamProvider):
         )
 
         async with httpx.AsyncClient(timeout=30.0) as client:
-            print(f"Payload: {payload}", "sending to", url)
             response = await client.post(url, headers=headers, json=payload)
             response.raise_for_status()
             invoice_data = response.json()

routstr/upstream/pricing_resolver.py

@@ -0,0 +1,203 @@
+"""Shared price/metadata resolution chain for upstream model discovery.
+
+Most OpenAI-compatible ``/models`` responses carry no pricing. Rather than let
+a provider fabricate one, this module resolves a model through decreasingly
+trustworthy sources — litellm's bundled cost map (curated list prices, mirrors
+provider docs), then the OpenRouter feed (resale prices, broader coverage) —
+and returns ``None`` when none of them know the model, so the caller can fail
+closed instead of inventing a number.
+
+Provider-native pricing (a gateway's own ``/models`` schema, e.g. Venice's
+``model_spec``) is authoritative and handled by the provider before this chain
+is consulted; only the shared fallback lives here so a later refactor can hoist
+it into the base provider unchanged.
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+
+
+@dataclass
+class ResolvedPricing:
+    """Per-token pricing plus whatever metadata the answering source carried.
+
+    Prices are USD per token. ``source`` records provenance
+    (``native``/``litellm``/``openrouter``/``unresolved``) so later work can
+    surface where each price came from.
+    """
+
+    prompt: float
+    completion: float
+    context_length: int | None
+    source: str
+    modality: str | None = None
+    max_completion_tokens: int | None = None
+    input_cache_read: float = 0.0
+    input_cache_write: float = 0.0
+    input_modalities: list[str] = field(default_factory=lambda: ["text"])
+    output_modalities: list[str] = field(default_factory=lambda: ["text"])
+    tokenizer: str = "unknown"
+    instruct_type: str | None = None
+    is_moderated: bool | None = None
+
+
+def estimate_context_length(model_id: str) -> int:
+    """Best-effort context window from a model id when no source reports one.
+
+    The last rung of the fallback chain, reached only for a model whose price
+    resolved but whose context did not (or that imported disabled). Context is
+    not a billing input, so a rough id-based guess is acceptable here where a
+    guessed *price* never would be.
+    """
+    lowered = model_id.lower()
+    if any(pattern in lowered for pattern in ["32k", "32000"]):
+        return 32768
+    if any(pattern in lowered for pattern in ["16k", "16000"]):
+        return 16384
+    if any(pattern in lowered for pattern in ["8k", "8000"]):
+        return 8192
+    if "gpt-4" in lowered:
+        return 8192
+    if "claude" in lowered:
+        return 200000
+    return 4096
+
+
+def _as_float(value: object) -> float | None:
+    """OpenRouter reports prices as strings; coerce, ``None`` if unparseable."""
+    try:
+        return float(value)  # type: ignore[arg-type]
+    except (TypeError, ValueError):
+        return None
+
+
+def _as_int(value: object) -> int | None:
+    """Coerce an already-numeric token count to ``int``, else ``None``."""
+    return int(value) if isinstance(value, (int, float)) else None
+
+
+def _from_litellm(model_id: str) -> ResolvedPricing | None:
+    # Lazy import so the resolver stays import-light and shares the exact
+    # lookup semantics used by cache-rate backfill.
+    from ..payment.models import litellm_cost_entry
+
+    info = litellm_cost_entry(model_id)
+    if info is None:
+        return None
+
+    prompt = info.get("input_cost_per_token")
+    completion = info.get("output_cost_per_token")
+    if not isinstance(prompt, (int, float)) or not isinstance(completion, (int, float)):
+        return None
+    # A both-zero entry is litellm listing a model without a real price (free
+    # moderation/rerank tiers do this) — treating 0/0 as resolved would serve
+    # the model for free. Reject it (and any negative) so the caller falls
+    # through, mirroring async_fetch_openrouter_models' _has_valid_pricing.
+    if prompt < 0 or completion < 0 or (prompt == 0 and completion == 0):
+        return None
+
+    input_modalities = ["text"]
+    if info.get("supports_vision"):
+        input_modalities.append("image")
+
+    return ResolvedPricing(
+        prompt=float(prompt),
+        completion=float(completion),
+        # max_input_tokens is the context window; max_tokens is litellm's
+        # completion cap (it tracks max_output_tokens for ~94% of models), so
+        # it is never a context source. A missing window falls to the id-based
+        # estimate downstream rather than borrowing the output cap.
+        context_length=_as_int(info.get("max_input_tokens")),
+        source="litellm",
+        max_completion_tokens=_as_int(info.get("max_output_tokens")),
+        input_cache_read=float(info.get("cache_read_input_token_cost") or 0.0),
+        input_cache_write=float(info.get("cache_creation_input_token_cost") or 0.0),
+        input_modalities=input_modalities,
+    )
+
+
+def _match_openrouter(model_id: str, feed: list[dict]) -> dict | None:
+    """Find ``model_id`` in the OpenRouter feed, exact id before bare tail.
+
+    Bare-tail matching (``deepseek-chat`` ↔ ``deepseek/deepseek-chat``) is a
+    looser, lower-trust match — OpenRouter fans a model out across resellers —
+    so an exact id match always wins first. When several entries share the bare
+    tail, the one with the highest *combined* (prompt + completion) per-token
+    cost wins: the choice must be deterministic (not feed-order-dependent) and
+    money-safe whichever way traffic leans, since undercharging is the hazard.
+    Ranking on prompt alone could pick an entry that is cheap on input but dear
+    on output. The live feed has no such collisions today; this only governs
+    the latent case.
+    """
+    bare = model_id.split("/", 1)[-1]
+    exact = next((m for m in feed if m.get("id") == model_id), None)
+    if exact is not None:
+        return exact
+    matches = [m for m in feed if m.get("id", "").split("/", 1)[-1] == bare]
+    if not matches:
+        return None
+
+    def _combined_cost(m: dict) -> float:
+        pricing = m.get("pricing", {})
+        return (_as_float(pricing.get("prompt")) or 0.0) + (
+            _as_float(pricing.get("completion")) or 0.0
+        )
+
+    return max(matches, key=_combined_cost)
+
+
+def _from_openrouter(model_id: str, feed: list[dict]) -> ResolvedPricing | None:
+    entry = _match_openrouter(model_id, feed)
+    if entry is None:
+        return None
+
+    pricing = entry.get("pricing", {})
+    prompt = _as_float(pricing.get("prompt"))
+    completion = _as_float(pricing.get("completion"))
+    if prompt is None or completion is None:
+        return None
+
+    architecture = entry.get("architecture", {})
+    top_provider = entry.get("top_provider", {})
+
+    return ResolvedPricing(
+        prompt=prompt,
+        completion=completion,
+        context_length=_as_int(entry.get("context_length")),
+        source="openrouter",
+        modality=architecture.get("modality"),
+        max_completion_tokens=_as_int(top_provider.get("max_completion_tokens")),
+        input_cache_read=_as_float(pricing.get("input_cache_read")) or 0.0,
+        input_cache_write=_as_float(pricing.get("input_cache_write")) or 0.0,
+        input_modalities=architecture.get("input_modalities") or ["text"],
+        output_modalities=architecture.get("output_modalities") or ["text"],
+        tokenizer=architecture.get("tokenizer") or "unknown",
+        instruct_type=architecture.get("instruct_type"),
+        is_moderated=top_provider.get("is_moderated"),
+    )
+
+
+class FallbackPricingResolver:
+    """Resolves models via litellm → OpenRouter for one discovery pass.
+
+    The OpenRouter catalog is fetched at most once and only when a model
+    actually misses litellm, so a provider full of litellm-known models never
+    touches the network. Instantiate one per ``fetch_models`` call.
+    """
+
+    def __init__(self) -> None:
+        self._openrouter_feed: list[dict] | None = None
+
+    async def resolve(self, model_id: str) -> ResolvedPricing | None:
+        """Resolve ``model_id``; ``None`` if no source knows it."""
+        resolved = _from_litellm(model_id)
+        if resolved is not None:
+            return resolved
+
+        if self._openrouter_feed is None:
+            # Lazy import so tests can patch the feed at its source.
+            from ..payment.models import async_fetch_openrouter_models
+
+            self._openrouter_feed = await async_fetch_openrouter_models()
+        return _from_openrouter(model_id, self._openrouter_feed)

routstr/upstream/rate_limit.py

@@ -0,0 +1,136 @@
+"""Detection and parsing of upstream provider rate-limit errors.
+
+Upstream OpenAI-compatible providers signal rate limits via HTTP 429 and/or a
+human-readable message such as::
+
+    Rate limit reached for gpt-5.5-2026-04-23 (for limit gpt-5.5) in organization
+    org-XXXX on tokens per min (TPM): Limit 180000000, Used 180000000,
+    Requested 8929. Please try again in 2ms.
+
+This module classifies those failures into a stable :data:`UPSTREAM_RATE_LIMIT`
+code and extracts useful debugging fields. All retained text is redacted of
+organization IDs first.
+"""
+
+from __future__ import annotations
+
+import re
+from dataclasses import asdict, dataclass
+
+from ..core.redaction import redact_org_ids
+
+# Stable error code callers can switch on to distinguish upstream rate limits
+# from generic request failures. The literal value matches the identifier named
+# in issue #555 ("UPSTREAM_RATE_LIMIT") so the public API contract is exact.
+UPSTREAM_RATE_LIMIT = "UPSTREAM_RATE_LIMIT"
+
+# Message fragments that indicate a rate-limit even when the status code is not
+# 429 (some providers wrap it in a 400/500 envelope).
+_RATE_LIMIT_MARKERS = (
+    "rate limit reached",
+    "rate_limit_exceeded",
+    "rate limit exceeded",
+    "too many requests",
+)
+
+_MODEL_RE = re.compile(r"Rate limit reached for ([^\s(]+)", re.IGNORECASE)
+_LIMIT_NAME_RE = re.compile(r"\(for limit ([^)]+)\)", re.IGNORECASE)
+_METRIC_RE = re.compile(r"on ([a-z ]+\((?:TPM|RPM|TPD|RPD|IPM)\))", re.IGNORECASE)
+_LIMIT_RE = re.compile(r"Limit (\d+)", re.IGNORECASE)
+_USED_RE = re.compile(r"Used (\d+)", re.IGNORECASE)
+_REQUESTED_RE = re.compile(r"Requested (\d+)", re.IGNORECASE)
+_RETRY_RE = re.compile(r"try again in ([\d.]+)\s*(ms|s)", re.IGNORECASE)
+
+
+@dataclass
+class RateLimitInfo:
+    """Structured, redaction-safe view of an upstream rate-limit error."""
+
+    code: str
+    message: str
+    model: str | None = None
+    limit_name: str | None = None
+    metric: str | None = None
+    limit: int | None = None
+    used: int | None = None
+    requested: int | None = None
+    retry_after_seconds: float | None = None
+
+    def as_details(self) -> dict[str, object]:
+        """Return a JSON-serialisable dict for embedding in an error envelope."""
+        return {k: v for k, v in asdict(self).items() if v is not None}
+
+
+def _looks_like_rate_limit(status_code: int, message: str) -> bool:
+    if status_code == 429:
+        return True
+    lowered = message.lower()
+    return any(marker in lowered for marker in _RATE_LIMIT_MARKERS)
+
+
+def _parse_retry_after_header(headers: dict[str, str] | None) -> float | None:
+    """Parse a ``Retry-After`` header (delta-seconds form) into seconds."""
+    if not headers:
+        return None
+    raw = headers.get("retry-after") or headers.get("Retry-After")
+    if raw is None:
+        return None
+    try:
+        return float(str(raw).strip())
+    except (TypeError, ValueError):
+        return None
+
+
+def _int_or_none(match: re.Match[str] | None) -> int | None:
+    if match is None:
+        return None
+    try:
+        return int(match.group(1))
+    except (TypeError, ValueError):
+        return None
+
+
+def classify_rate_limit(
+    status_code: int,
+    message: str,
+    headers: dict[str, str] | None = None,
+) -> RateLimitInfo | None:
+    """Classify an upstream error as a rate-limit and extract its fields.
+
+    Args:
+        status_code: HTTP status code from the upstream response.
+        message: Upstream error message (may contain sensitive identifiers).
+        headers: Optional upstream response headers, used for ``Retry-After``.
+
+    Returns:
+        A :class:`RateLimitInfo` when the error is a rate-limit, else ``None``.
+    """
+    message = message or ""
+    if not _looks_like_rate_limit(status_code, message):
+        return None
+
+    redacted = redact_org_ids(message)
+
+    model_match = _MODEL_RE.search(redacted)
+    metric_match = _METRIC_RE.search(redacted)
+
+    retry_after = _parse_retry_after_header(headers)
+    if retry_after is None:
+        retry_match = _RETRY_RE.search(redacted)
+        if retry_match is not None:
+            value = float(retry_match.group(1))
+            retry_after = value / 1000.0 if retry_match.group(2).lower() == "ms" else value
+
+    limit_name_match = _LIMIT_NAME_RE.search(redacted)
+
+    return RateLimitInfo(
+        code=UPSTREAM_RATE_LIMIT,
+        message=redacted,
+        model=model_match.group(1) if model_match else None,
+        limit_name=limit_name_match.group(1).strip() if limit_name_match else None,
+        metric=metric_match.group(1).strip() if metric_match else None,
+        limit=_int_or_none(_LIMIT_RE.search(redacted)),
+        used=_int_or_none(_USED_RE.search(redacted)),
+        requested=_int_or_none(_REQUESTED_RE.search(redacted)),
+        retry_after_seconds=retry_after,
+    )

routstr/upstream/request_correction.py

@@ -0,0 +1,142 @@
+"""Reactive request-correction layer.
+
+When an upstream rejects a request with a recoverable 4xx error, this layer
+tries to *fix* the request body and let the caller retry the same upstream
+instead of failing outright. It is provider-agnostic: correctors key off the
+upstream's own error wording, so the same recovery works across every provider.
+
+The layer is a small pipeline of :data:`Corrector` callables. Each corrector
+inspects the parsed request body and the upstream error message and either
+returns a corrected body (plus a short label identifying the fix) or declines
+by returning ``None``. Adding a new reactive fix means writing one corrector
+and adding it to :data:`DEFAULT_CORRECTORS` — no changes to the proxy loop.
+
+All corrections are immutable: a corrector never mutates the body it is given,
+it returns a new ``dict``. The proxy threads an ``applied`` set of fix labels
+through retries so each distinct fix is applied at most once, guaranteeing the
+retry loop always terminates.
+"""
+
+from __future__ import annotations
+
+import json
+import re
+from collections.abc import Callable, Sequence
+from dataclasses import dataclass
+
+from fastapi.responses import Response
+
+from ..core import get_logger
+
+logger = get_logger(__name__)
+
+
+# Matches upstream error text that names a single rejected request parameter,
+# e.g. "`temperature` is deprecated for this model." or
+# "parameter 'top_p' is not supported". Keys off the upstream's own wording so
+# a 400 about an unsupported sampling/option field can be recovered by stripping
+# that field and retrying the same upstream.
+_UNSUPPORTED_PARAM_RE = re.compile(
+    r"[`'\"]?(?P<param>[a-zA-Z_][a-zA-Z0-9_]*)[`'\"]?\s+is\s+"
+    r"(?:deprecated|not\s+supported|unsupported|no\s+longer\s+supported)",
+    re.IGNORECASE,
+)
+
+
+# A corrector inspects the parsed request body and the upstream error message
+# and returns ``(new_body_dict, label)`` for a fix it can apply, or ``None`` to
+# decline. ``label`` identifies the fix so it is applied at most once per request.
+Corrector = Callable[[dict, str], "tuple[dict, str] | None"]
+
+
+@dataclass(frozen=True)
+class Correction:
+    """A successful request correction ready to retry.
+
+    ``body`` is the corrected JSON body (encoded), ``label`` identifies the fix
+    that was applied (e.g. the stripped param name) so the caller can guard
+    against applying the same fix twice.
+    """
+
+    body: bytes
+    label: str
+
+
+def extract_error_message(response: Response) -> str:
+    """Best-effort extraction of an error message string from a proxy Response."""
+    body_bytes = getattr(response, "body", None)
+    if not body_bytes:
+        return ""
+    try:
+        data = json.loads(body_bytes)
+    except Exception:
+        return body_bytes.decode("utf-8", errors="ignore")[:500]
+    if isinstance(data, dict):
+        err = data.get("error")
+        if isinstance(err, dict):
+            msg = err.get("message") or err.get("detail")
+            if isinstance(msg, str):
+                return msg
+        elif isinstance(err, str):
+            return err
+        if isinstance(data.get("message"), str):
+            return data["message"]
+    return ""
+
+
+def strip_unsupported_param(
+    body: dict, error_message: str
+) -> tuple[dict, str] | None:
+    """Drop a top-level param the upstream named as unsupported/deprecated.
+
+    Returns ``(new_body, param)`` (a new dict, original untouched) when the
+    error names a top-level param present in the body, otherwise ``None``.
+    """
+    match = _UNSUPPORTED_PARAM_RE.search(error_message)
+    if not match:
+        return None
+    param = match.group("param")
+    if param not in body:
+        return None
+    new_body = {k: v for k, v in body.items() if k != param}
+    return new_body, param
+
+
+# Ordered pipeline of correctors tried on each recoverable rejection.
+DEFAULT_CORRECTORS: tuple[Corrector, ...] = (strip_unsupported_param,)
+
+
+def correct_request(
+    request_body: bytes,
+    error_message: str,
+    applied: set[str],
+    correctors: Sequence[Corrector] = DEFAULT_CORRECTORS,
+) -> Correction | None:
+    """Try to correct a rejected request body so it can be retried.
+
+    Runs each corrector in order against the parsed body and ``error_message``.
+    The first corrector that proposes a fix whose ``label`` is not already in
+    ``applied`` wins; its result is returned as a :class:`Correction`. Returns
+    ``None`` when nothing parses, nothing matches, or every proposed fix was
+    already applied — the caller then treats the response as a normal failure.
+
+    ``applied`` is read-only here; the caller records the returned ``label`` to
+    bound retries and guarantee forward progress.
+    """
+    if not request_body or not error_message:
+        return None
+    try:
+        data = json.loads(request_body)
+    except Exception:
+        return None
+    if not isinstance(data, dict):
+        return None
+    for corrector in correctors:
+        result = corrector(data, error_message)
+        if result is None:
+            continue
+        new_body, label = result
+        if label in applied:
+            continue
+        return Correction(body=json.dumps(new_body).encode(), label=label)
+    return None

routstr/upstream/routstr.py

@@ -0,0 +1,180 @@
+from typing import TYPE_CHECKING, Any
+
+import httpx
+
+from ..core import get_logger
+from ..payment.models import Model
+from .base import BaseUpstreamProvider
+
+if TYPE_CHECKING:
+    from ..core.db import UpstreamProviderRow
+
+logger = get_logger(__name__)
+
+
+class RoutstrUpstreamProvider(BaseUpstreamProvider):
+    """Upstream provider for communicating with another Routstr instance."""
+
+    provider_type = "routstr"
+    default_base_url = None
+    platform_url = None
+    # Upstream Routstr nodes serve `/v1/messages` natively, so forward the
+    # request as-is instead of round-tripping through litellm's
+    # Anthropic→OpenAI translator.
+    supports_anthropic_messages = True
+
+    def __init__(
+        self,
+        base_url: str,
+        api_key: str,
+        provider_fee: float = 1.01,
+        provider_settings: dict | None = None,
+    ):
+        """Initialize Routstr provider.
+
+        Args:
+            base_url: Base URL of the upstream Routstr instance
+            api_key: API key for the upstream Routstr instance
+            provider_fee: Provider fee multiplier
+            provider_settings: Provider-specific settings (auto-topup, etc.)
+        """
+        # Ensure base_url doesn't end with /v1 as BaseUpstreamProvider appends it if needed
+        # but Routstr paths are usually absolute from base.
+        super().__init__(
+            base_url=base_url.rstrip("/"),
+            api_key=api_key,
+            provider_fee=provider_fee,
+        )
+        self.settings = provider_settings or {}
+
+    def normalize_request_path(
+        self, path: str, model_obj: "Model | None" = None
+    ) -> str:
+        """Preserve the ``v1/`` prefix when forwarding to an upstream Routstr.
+        """
+        return path.lstrip("/")
+
+    @classmethod
+    def _build_from_row(
+        cls, provider_row: "UpstreamProviderRow"
+    ) -> "RoutstrUpstreamProvider":
+        import json
+
+        settings = {}
+        if provider_row.provider_settings:
+            try:
+                settings = json.loads(provider_row.provider_settings)
+            except Exception:
+                pass
+
+        return cls(
+            base_url=provider_row.base_url,
+            api_key=provider_row.api_key,
+            provider_fee=provider_row.provider_fee,
+            provider_settings=settings,
+        )
+
+    @classmethod
+    def get_provider_metadata(cls) -> dict[str, object]:
+        return {
+            "id": cls.provider_type,
+            "name": "Routstr Node",
+            "default_base_url": "",
+            "fixed_base_url": False,
+            "platform_url": cls.platform_url,
+            "can_create_account": False,
+            "can_topup": True,
+            "can_show_balance": True,
+        }
+
+    async def get_balance(self) -> float | None:
+        """Fetch balance from the upstream Routstr node.
+
+        Returns:
+            Balance in satoshis, or None if failed
+        """
+        url = f"{self.base_url}/v1/balance/info"
+        headers = {"Authorization": f"Bearer {self.api_key}"} if self.api_key else {}
+
+        async with httpx.AsyncClient() as client:
+            try:
+                response = await client.get(url, headers=headers, timeout=10.0)
+                response.raise_for_status()
+                data = response.json()
+                # Routstr balance info usually contains 'balance' in msats or sats
+                # Check for msats and convert to sats
+                if "balance_msats" in data:
+                    return float(data["balance_msats"]) / 1000.0
+                return float(data.get("balance", 0))
+            except Exception as e:
+                logger.error(
+                    "Failed to fetch balance from upstream Routstr",
+                    extra={"url": url, "error": str(e)},
+                )
+                return None
+
+    async def topup(self, cashu_token: str) -> dict[str, Any]:
+        """Top up balance on the upstream Routstr node.
+
+        Args:
+            cashu_token: Cashu token to deposit
+
+        Returns:
+            Dict containing top-up result
+        """
+        url = f"{self.base_url}/v1/balance/topup"
+        headers = {"Authorization": f"Bearer {self.api_key}"}
+        payload = {"cashu_token": cashu_token}
+
+        async with httpx.AsyncClient() as client:
+            try:
+                response = await client.post(
+                    url, headers=headers, json=payload, timeout=30.0
+                )
+                response.raise_for_status()
+                return response.json()
+            except Exception as e:
+                logger.error(
+                    "Failed to topup upstream Routstr",
+                    extra={"url": url, "error": str(e)},
+                )
+                return {"error": str(e)}
+
+    async def fetch_models(self) -> list[Model]:
+        """Fetch models from the upstream Routstr node."""
+        url = f"{self.base_url}/v1/models"
+
+        async with httpx.AsyncClient() as client:
+            try:
+                response = await client.get(url, headers={}, timeout=15.0)
+                response.raise_for_status()
+                data = response.json()
+                models = data.get("data", [])
+                return [Model(**m) for m in models]
+            except Exception as e:
+                logger.error(
+                    "Failed to fetch models from upstream Routstr",
+                    extra={"url": url, "error": str(e)},
+                )
+                return []
+
+    async def refund_balance(self) -> dict[str, Any]:
+        """Request a refund from the upstream Routstr node.
+
+        Returns:
+            Dict containing refund result and token
+        """
+        url = f"{self.base_url}/v1/balance/refund"
+        headers = {"Authorization": f"Bearer {self.api_key}"}
+
+        async with httpx.AsyncClient() as client:
+            try:
+                response = await client.post(url, headers=headers, timeout=30.0)
+                response.raise_for_status()
+                return response.json()
+            except Exception as e:
+                logger.error(
+                    "Failed to request refund from upstream Routstr",
+                    extra={"url": url, "error": str(e)},
+                )
+                return {"error": str(e)}

routstr/upstream/tinfoil.py

@@ -0,0 +1,236 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
+import httpx
+from fastapi import Request
+from fastapi.responses import Response, StreamingResponse
+from pydantic.v1 import BaseModel
+
+from ..core.exceptions import UpstreamError
+from ..core.logging import get_logger
+from ..payment.models import Architecture, Model, Pricing
+from .base import BaseUpstreamProvider
+from .ehbp import (
+    _ENCLAVE_URL_HEADER,
+    _PROXY_ONLY_HEADERS,
+    _RESPONSE_USAGE_HEADER,
+    ConfidentialInferenceProfile,
+    EHBPForwardingTarget,
+)
+
+if TYPE_CHECKING:
+    from ..core.db import UpstreamProviderRow
+
+logger = get_logger(__name__)
+
+
+class TinfoilModelPricing(BaseModel):
+    inputTokenPricePer1M: float = 0.0
+    outputTokenPricePer1M: float = 0.0
+    requestPrice: float = 0.0
+
+
+class TinfoilModel(BaseModel):
+    id: str
+    context_window: int = 0
+    created: int = 0
+    multimodal: bool = False
+    reasoning: bool = False
+    tool_calling: bool = False
+    type: str = "chat"
+    pricing: TinfoilModelPricing = TinfoilModelPricing()
+    endpoints: list[str] = []
+
+
+class TinfoilUpstreamProvider(BaseUpstreamProvider):
+    """Direct upstream provider for the Tinfoil inference API.
+
+    Tinfoil hosts open-source models inside attested secure enclaves and exposes
+    an OpenAI-compatible API at ``https://inference.tinfoil.sh``. Request and
+    response bodies are encrypted end-to-end with EHBP (HPKE), so Routstr acts
+    as a blind relay: it forwards the opaque encrypted body, never sees
+    plaintext, and bills from the ``X-Tinfoil-Usage-Metrics`` header that
+    Tinfoil returns outside the encrypted body when
+    ``X-Tinfoil-Request-Usage-Metrics: true`` is set.
+    """
+
+    provider_type = "tinfoil"
+    default_base_url = "https://inference.tinfoil.sh"
+    platform_url = "https://docs.tinfoil.sh"
+    supports_ehbp = True
+    confidential_inference_profile = ConfidentialInferenceProfile(
+        usage_response_header=_RESPONSE_USAGE_HEADER,
+        client_target_url_header=_ENCLAVE_URL_HEADER,
+        allow_client_target_override=True,
+        proxy_only_headers=_PROXY_ONLY_HEADERS,
+    )
+
+    def __init__(self, api_key: str, provider_fee: float = 1.0):
+        super().__init__(
+            base_url=self.default_base_url,
+            api_key=api_key,
+            provider_fee=provider_fee,
+        )
+
+    @classmethod
+    def _build_from_row(
+        cls, provider_row: "UpstreamProviderRow"
+    ) -> "TinfoilUpstreamProvider":
+        return cls(
+            api_key=provider_row.api_key,
+            provider_fee=provider_row.provider_fee,
+        )
+
+    @classmethod
+    def get_provider_metadata(cls) -> dict[str, object]:
+        return {
+            "id": cls.provider_type,
+            "name": "Tinfoil",
+            "default_base_url": cls.default_base_url,
+            "fixed_base_url": True,
+            "platform_url": cls.platform_url,
+            "can_create_account": False,
+            "can_topup": False,
+            "can_show_balance": False,
+        }
+
+    def transform_model_name(self, model_id: str) -> str:
+        return model_id.removeprefix("tinfoil/")
+
+    def get_confidential_inference_profile(self) -> ConfidentialInferenceProfile:
+        return self.confidential_inference_profile
+
+    async def forward_get_request(
+        self,
+        request: Request,
+        path: str,
+        headers: dict,
+    ) -> Response | StreamingResponse:
+        """Handle Tinfoil-specific GET endpoints.
+
+        * ``/attestation`` (or ``/tee/attestation``): proxy to the Tinfoil ATC
+          (attestation bundle proxy) at ``https://atc.tinfoil.sh/attestation``.
+        * Other GETs: forward to the provider base URL
+          (``https://inference.tinfoil.sh``). ``X-Tinfoil-Enclave-Url`` is an
+          EHBP-only header used for encrypted POST requests and is not honored
+          for unencrypted GET requests.
+        """
+        clean_path = path.removeprefix("tee/").rstrip("/")
+        if clean_path == "attestation":
+            return await self._proxy_attestation(headers)
+        return await super().forward_get_request(request, path, headers)
+
+    async def _proxy_attestation(self, headers: dict) -> Response:
+        url = "https://atc.tinfoil.sh/attestation"
+        async with httpx.AsyncClient(
+            transport=httpx.AsyncHTTPTransport(retries=1),
+            timeout=30.0,
+        ) as client:
+            try:
+                resp = await client.get(
+                    url,
+                    headers={
+                        "Accept": headers.get("accept", "application/json"),
+                    },
+                )
+                response_headers = dict(resp.headers)
+                response_headers.pop("content-encoding", None)
+                response_headers.pop("content-length", None)
+                return Response(
+                    content=resp.content,
+                    status_code=resp.status_code,
+                    headers=response_headers,
+                )
+            except Exception as exc:
+                raise UpstreamError(
+                    f"Error fetching Tinfoil attestation: {type(exc).__name__}",
+                    status_code=502,
+                ) from exc
+
+    def get_ehbp_forwarding_target(
+        self, path: str, model_obj: Model
+    ) -> EHBPForwardingTarget:
+        """Return the Tinfoil enclave target for EHBP requests.
+
+        Requests usage metrics from the enclave so Routstr can bill exactly
+        without decrypting the response body. The actual forwarding URL is
+        overridden at dispatch time by ``X-Tinfoil-Enclave-Url`` when the SDK
+        sends it (see ``routstr/upstream/ehbp.py``).
+        """
+        return EHBPForwardingTarget(
+            url=f"{self.base_url.rstrip('/')}/{path.lstrip('/')}",
+            headers={"X-Tinfoil-Request-Usage-Metrics": "true"},
+            profile=self.confidential_inference_profile,
+        )
+
+    async def fetch_models(self) -> list[Model]:
+        """Fetch models from the public Tinfoil models endpoint.
+
+        ``GET /v1/models`` is unauthenticated and returns all available models
+        with their pricing in USD per 1M tokens.
+        """
+        url = f"{self.base_url}/v1/models"
+        try:
+            async with httpx.AsyncClient(timeout=30.0) as client:
+                response = await client.get(url)
+                response.raise_for_status()
+                data = response.json()
+                models_data = data.get("data", [])
+
+                models: list[Model] = []
+                for model_data in models_data:
+                    try:
+                        tf = TinfoilModel.parse_obj(model_data)
+                        input_price = tf.pricing.inputTokenPricePer1M
+                        output_price = tf.pricing.outputTokenPricePer1M
+                        request_price = tf.pricing.requestPrice
+
+                        modality = "text->text"
+                        input_modalities = ["text"]
+                        output_modalities = ["text"]
+                        if tf.multimodal:
+                            modality = "text->text+image"
+                            input_modalities = ["text", "image"]
+
+                        models.append(
+                            Model(
+                                id=tf.id,
+                                name=tf.id,
+                                created=tf.created,
+                                description=f"Tinfoil {tf.type} model",
+                                context_length=tf.context_window,
+                                architecture=Architecture(
+                                    modality=modality,
+                                    input_modalities=input_modalities,
+                                    output_modalities=output_modalities,
+                                    tokenizer="Unknown",
+                                    instruct_type=None,
+                                ),
+                                pricing=Pricing(
+                                    prompt=input_price / 1_000_000,
+                                    completion=output_price / 1_000_000,
+                                    request=request_price,
+                                    image=0.0,
+                                    web_search=0.0,
+                                    internal_reasoning=0.0,
+                                ),
+                            )
+                        )
+                    except Exception as e:
+                        logger.warning(
+                            "Failed to parse Tinfoil model",
+                            extra={
+                                "model_id": model_data.get("id", "unknown"),
+                                "error": str(e),
+                                "error_type": type(e).__name__,
+                            },
+                        )
+
+                return models
+        except Exception as e:
+            logger.error(
+                "Error fetching models from Tinfoil",
+                extra={"error": str(e), "error_type": type(e).__name__},
+            )
+            return []

routstr/upstream/tinfoil_trailer.py

@@ -0,0 +1,189 @@
+"""h11-based HTTP client for EHBP requests that captures HTTP trailers.
+
+httpx/httpcore silently discard HTTP trailers during chunked transfer
+decoding. Tinfoil returns ``X-Tinfoil-Usage-Metrics`` as a trailer on
+streaming responses, so we need a lower-level HTTP client that preserves
+trailers from the h11 ``EndOfMessage`` event.
+
+Because EHBP response bodies are opaque encrypted blobs, buffering the full
+response is acceptable — the client decrypts the complete body regardless of
+whether it arrived streamed or buffered.
+"""
+
+from __future__ import annotations
+
+import asyncio
+import ssl
+from dataclasses import dataclass, field
+from urllib.parse import urlsplit
+
+import h11
+
+from ..core import get_logger
+
+logger = get_logger(__name__)
+
+_READ_BUFSIZE = 65536
+_DEFAULT_TIMEOUT_SECONDS = 30.0
+_DEFAULT_CLOSE_TIMEOUT_SECONDS = 1.0
+_DEFAULT_MAX_RESPONSE_BYTES = 25 * 1024 * 1024
+_HOP_BY_HOP_HEADERS = {
+    "connection",
+    "keep-alive",
+    "proxy-authenticate",
+    "proxy-authorization",
+    "te",
+    "trailer",
+    "transfer-encoding",
+    "upgrade",
+}
+
+
+@dataclass
+class TrailerResponse:
+    """Buffered HTTP response with optional trailer headers."""
+
+    status_code: int
+    headers: list[tuple[str, str]]
+    body: bytes
+    trailers: list[tuple[str, str]] = field(default_factory=list)
+
+
+def _get_header(headers: list[tuple[str, str]], name: str) -> str | None:
+    name_lower = name.lower()
+    for k, v in headers:
+        if k.lower() == name_lower:
+            return v
+    return None
+
+
+def _strip_hop_by_hop_headers(headers: dict[str, str]) -> dict[str, str]:
+    """Remove connection-specific headers before serializing a new request."""
+    connection_tokens: set[str] = set()
+    for key, value in headers.items():
+        if key.lower() == "connection":
+            connection_tokens.update(
+                token.strip().lower() for token in value.split(",") if token.strip()
+            )
+
+    excluded = _HOP_BY_HOP_HEADERS | connection_tokens
+    return {key: value for key, value in headers.items() if key.lower() not in excluded}
+
+
+async def forward_with_trailer(
+    *,
+    method: str,
+    url: str,
+    headers: dict[str, str],
+    body: bytes,
+    timeout_seconds: float = _DEFAULT_TIMEOUT_SECONDS,
+    max_response_bytes: int = _DEFAULT_MAX_RESPONSE_BYTES,
+    close_timeout_seconds: float = _DEFAULT_CLOSE_TIMEOUT_SECONDS,
+) -> TrailerResponse:
+    """Send an HTTP/1.1 request via h11 and capture HTTP trailers.
+
+    Returns a :class:`TrailerResponse` with the full buffered body and any
+    trailer headers from the ``EndOfMessage`` event.
+    """
+    parsed = urlsplit(url)
+    host = parsed.hostname
+    if not host:
+        raise ValueError(f"Invalid URL (no hostname): {url}")
+    port = parsed.port or 443
+    path = parsed.path or "/"
+    if parsed.query:
+        path = f"{path}?{parsed.query}"
+
+    # FastAPI has already decoded the incoming request body. Do not carry the
+    # original connection's framing or other hop-by-hop metadata into the new
+    # upstream connection.
+    headers = _strip_hop_by_hop_headers(headers)
+
+    ssl_ctx = ssl.create_default_context()
+    reader, writer = await asyncio.wait_for(
+        asyncio.open_connection(host, port, ssl=ssl_ctx),
+        timeout=timeout_seconds,
+    )
+
+    try:
+        # Build HTTP/1.1 request
+        header_lines = [f"{method} {path} HTTP/1.1"]
+        has_host = any(k.lower() == "host" for k in headers)
+        if not has_host:
+            header_lines.append(f"Host: {host}")
+        header_lines.append("Connection: close")
+
+        for key, value in headers.items():
+            if key.lower() == "host":
+                continue
+            header_lines.append(f"{key}: {value}")
+
+        if body and not any(k.lower() == "content-length" for k in headers):
+            header_lines.append(f"Content-Length: {len(body)}")
+
+        request_data = "\r\n".join(header_lines).encode() + b"\r\n\r\n"
+        if body:
+            request_data += body
+
+        writer.write(request_data)
+        await asyncio.wait_for(writer.drain(), timeout=timeout_seconds)
+
+        # Parse response with h11
+        conn = h11.Connection(h11.CLIENT)
+        status_code = 0
+        resp_headers: list[tuple[str, str]] = []
+        body_chunks: list[bytes] = []
+        body_size = 0
+        trailers: list[tuple[str, str]] = []
+
+        while True:
+            event = conn.next_event()
+
+            if event is h11.NEED_DATA:
+                data = await asyncio.wait_for(
+                    reader.read(_READ_BUFSIZE),
+                    timeout=timeout_seconds,
+                )
+                conn.receive_data(data if data else b"")
+                continue
+
+            if isinstance(event, h11.Response):
+                status_code = event.status_code
+                resp_headers = [(k.decode(), v.decode()) for k, v in event.headers]
+
+            elif isinstance(event, h11.Data):
+                body_size += len(event.data)
+                if body_size > max_response_bytes:
+                    raise ValueError(
+                        f"EHBP response exceeded {max_response_bytes} bytes"
+                    )
+                body_chunks.append(event.data)
+
+            elif isinstance(event, h11.EndOfMessage):
+                for k, v in event.headers:
+                    trailers.append((k.decode(), v.decode()))
+                break
+
+            elif isinstance(event, h11.PAUSED):
+                # Shouldn't happen for simple request/response, but break safely
+                logger.warning("h11 PAUSED event during EHBP response parsing")
+                break
+
+            elif isinstance(event, h11.ConnectionClosed):
+                break
+
+        return TrailerResponse(
+            status_code=status_code,
+            headers=resp_headers,
+            body=b"".join(body_chunks),
+            trailers=trailers,
+        )
+    finally:
+        writer.close()
+        if close_timeout_seconds > 0:
+            try:
+                await asyncio.wait_for(
+                    writer.wait_closed(), timeout=close_timeout_seconds
+                )
+            except Exception:
+                pass

routstr/upstream/xai.py

@@ -13,6 +13,7 @@ class XAIUpstreamProvider(BaseUpstreamProvider):
     provider_type = "x-ai"
     default_base_url = "https://api.x.ai/v1"
     platform_url = "https://console.x.ai/"
+    litellm_provider_prefix = "xai/"
 
     def __init__(self, api_key: str, provider_fee: float = 1.01):
         super().__init__(
@@ -20,7 +21,7 @@ class XAIUpstreamProvider(BaseUpstreamProvider):
         )
 
     @classmethod
-    def from_db_row(cls, provider_row: "UpstreamProviderRow") -> "XAIUpstreamProvider":
+    def _build_from_row(cls, provider_row: "UpstreamProviderRow") -> "XAIUpstreamProvider":
         return cls(
             api_key=provider_row.api_key,
             provider_fee=provider_row.provider_fee,

tests/integration/conftest.py

@@ -174,17 +174,20 @@ class TestmintWallet:
         """Fallback method to create a basic test token"""
         import base64
         import json
-        import random
-        import time
+        import secrets
 
-        unique_id = int(time.time() * 1000000) + random.randint(1000, 9999)
+        # Use a cryptographically random id/secret so every minted token is
+        # guaranteed unique. Time/PRNG-based ids can collide on hosts with
+        # coarse clock resolution, producing byte-identical tokens that hash to
+        # the same api_key and silently dedupe (flaky concurrency tests).
+        unique_id = secrets.token_hex(16)
         token_data = {
             "token": [
                 {
                     "mint": self.mint_url,
                     "proofs": [
                         {
-                            "id": f"009a1f293253e41e{unique_id % 100000000:08d}",
+                            "id": f"009a1f293253e41e{unique_id[:8]}",
                             "amount": amount,
                             "secret": f"test-secret-{amount}-{unique_id}",
                             "C": "02194603ffa36356f4a56b7df9371fc3192472351453ec7398b8da8117e7c3e104",
@@ -380,6 +383,13 @@ async def integration_session(
         yield session
 
 
+@pytest_asyncio.fixture
+async def patched_db_engine(integration_engine: Any) -> AsyncGenerator[None, None]:
+    """Patch the global db engine so create_session() uses the test engine."""
+    with patch("routstr.core.db.engine", integration_engine):
+        yield
+
+
 class DatabaseSnapshot:
     """Utility to capture and compare database states"""
 

tests/integration/test_admin_model_cache_pricing.py

@@ -0,0 +1,229 @@
+import json
+from datetime import datetime, timedelta, timezone
+from unittest.mock import patch
+
+import pytest
+from httpx import AsyncClient
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.admin import admin_sessions
+from routstr.core.db import ModelRow, UpstreamProviderRow
+from routstr.payment.cost_calculation import CostData, calculate_cost
+from routstr.proxy import get_model_instance, reinitialize_upstreams
+
+
+def _admin_headers() -> dict[str, str]:
+    token = "test-admin-cache-pricing-token"
+    admin_sessions[token] = int(
+        (datetime.now(timezone.utc) + timedelta(minutes=5)).timestamp()
+    )
+    return {"Authorization": f"Bearer {token}"}
+
+
+def _model_payload(
+    provider_id: int,
+    *,
+    cache_read: float,
+    cache_write: float,
+) -> dict[str, object]:
+    return {
+        "id": "custom-cache-model",
+        "name": "Custom Cache Model",
+        "description": "custom model with explicit cache pricing",
+        "created": 0,
+        "context_length": 128000,
+        "architecture": {
+            "modality": "text",
+            "input_modalities": ["text"],
+            "output_modalities": ["text"],
+            "tokenizer": "unknown",
+            "instruct_type": None,
+        },
+        "pricing": {
+            "prompt": 1.4e-7,
+            "completion": 2.8e-7,
+            "input_cache_read": cache_read,
+            "input_cache_write": cache_write,
+            "request": 0.0,
+            "image": 0.0,
+            "web_search": 0.0,
+            "internal_reasoning": 0.0,
+        },
+        "per_request_limits": None,
+        "top_provider": None,
+        "upstream_provider_id": provider_id,
+        "canonical_slug": None,
+        "alias_ids": [],
+        "enabled": True,
+        "forwarded_model_id": "custom-cache-model",
+    }
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_admin_provider_model_api_persists_cache_pricing_on_create_and_update(
+    integration_client: AsyncClient,
+    integration_session: AsyncSession,
+) -> None:
+    provider = UpstreamProviderRow(
+        provider_type="generic",
+        base_url="https://custom-upstream.example/v1",
+        api_key="test-key",
+        provider_fee=1.0,
+    )
+    integration_session.add(provider)
+    await integration_session.commit()
+    await integration_session.refresh(provider)
+    assert provider.id is not None
+    await reinitialize_upstreams()
+
+    headers = _admin_headers()
+    create_payload = _model_payload(
+        provider.id,
+        cache_read=2.8e-9,
+        cache_write=3.5e-9,
+    )
+
+    with patch("routstr.payment.models.sats_usd_price", return_value=1e-6):
+        create_response = await integration_client.post(
+            f"/admin/api/upstream-providers/{provider.id}/models",
+            headers=headers,
+            json=create_payload,
+        )
+
+    assert create_response.status_code == 200
+    create_body = create_response.json()
+    assert create_body["pricing"]["input_cache_read"] == pytest.approx(2.8e-9)
+    assert create_body["pricing"]["input_cache_write"] == pytest.approx(3.5e-9)
+
+    row = await integration_session.get(ModelRow, ("custom-cache-model", provider.id))
+    assert row is not None
+    stored_pricing = json.loads(row.pricing)
+    assert stored_pricing["input_cache_read"] == pytest.approx(2.8e-9)
+    assert stored_pricing["input_cache_write"] == pytest.approx(3.5e-9)
+
+    update_payload = _model_payload(
+        provider.id,
+        cache_read=1.25e-9,
+        cache_write=4.5e-9,
+    )
+
+    with patch("routstr.payment.models.sats_usd_price", return_value=1e-6):
+        update_response = await integration_client.post(
+            f"/admin/api/upstream-providers/{provider.id}/models",
+            headers=headers,
+            json=update_payload,
+        )
+
+    assert update_response.status_code == 200
+    update_body = update_response.json()
+    assert update_body["pricing"]["input_cache_read"] == pytest.approx(1.25e-9)
+    assert update_body["pricing"]["input_cache_write"] == pytest.approx(4.5e-9)
+
+    await integration_session.refresh(row)
+    updated_pricing = json.loads(row.pricing)
+    assert updated_pricing["input_cache_read"] == pytest.approx(1.25e-9)
+    assert updated_pricing["input_cache_write"] == pytest.approx(4.5e-9)
+
+    model = get_model_instance("custom-cache-model")
+    assert model is not None
+    assert model.sats_pricing is not None
+    assert model.sats_pricing.input_cache_read == pytest.approx(0.00125)
+    assert model.sats_pricing.input_cache_write == pytest.approx(0.0045)
+
+    with patch("routstr.payment.cost_calculation.sats_usd_price", return_value=1e-6):
+        cost = await calculate_cost(
+            {
+                "model": "custom-cache-model",
+                "usage": {
+                    "prompt_tokens": 1000,
+                    "completion_tokens": 100,
+                    "prompt_tokens_details": {"cached_tokens": 800},
+                },
+            },
+            max_cost=1_000_000,
+        )
+
+    assert isinstance(cost, CostData)
+    assert cost.input_tokens == 200
+    assert cost.cache_read_input_tokens == 800
+    assert cost.cache_read_msats == 1000
+    assert cost.output_msats == 28000
+    assert cost.input_msats == 29000
+    assert cost.total_msats == 57000
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_upstream_response_cost_uses_model_cache_pricing(
+    integration_session: AsyncSession,
+    patched_db_engine: None,
+) -> None:
+    """A model's configured cache price must discount upstream cached-token usage."""
+    provider = UpstreamProviderRow(
+        provider_type="generic",
+        base_url="https://cache-priced-upstream.example/v1",
+        api_key="test-key",
+        provider_fee=1.0,
+    )
+    integration_session.add(provider)
+    await integration_session.commit()
+    await integration_session.refresh(provider)
+    assert provider.id is not None
+
+    row = ModelRow(
+        id="cache-priced-model",
+        name="Cache Priced Model",
+        description="model seeded with explicit cache pricing",
+        created=0,
+        context_length=128000,
+        architecture=json.dumps(
+            {
+                "modality": "text",
+                "input_modalities": ["text"],
+                "output_modalities": ["text"],
+                "tokenizer": "unknown",
+                "instruct_type": None,
+            }
+        ),
+        pricing=json.dumps(
+            {
+                "prompt": 1.4e-7,
+                "completion": 2.8e-7,
+                "input_cache_read": 1.25e-9,
+                "input_cache_write": 4.5e-9,
+            }
+        ),
+        upstream_provider_id=provider.id,
+        enabled=True,
+        forwarded_model_id="cache-priced-model",
+    )
+    integration_session.add(row)
+    await integration_session.commit()
+
+    with patch("routstr.payment.models.sats_usd_price", return_value=1e-6):
+        await reinitialize_upstreams()
+
+    with patch("routstr.payment.cost_calculation.sats_usd_price", return_value=1e-6):
+        cost = await calculate_cost(
+            {
+                "model": "cache-priced-model",
+                "usage": {
+                    "prompt_tokens": 1000,
+                    "completion_tokens": 100,
+                    "prompt_tokens_details": {"cached_tokens": 800},
+                },
+            },
+            max_cost=1_000_000,
+        )
+
+    assert isinstance(cost, CostData)
+    # prompt 1.4e-7 USD/token -> 0.14 sats/token -> 140 msats/token
+    # cache 1.25e-9 USD/token -> 0.00125 sats/token -> 1.25 msats/token
+    # completion 2.8e-7 USD/token -> 0.28 sats/token -> 280 msats/token
+    assert cost.input_tokens == 200
+    assert cost.cache_read_input_tokens == 800
+    assert cost.cache_read_msats == 1000
+    assert cost.input_msats == 29000
+    assert cost.output_msats == 28000
+    assert cost.total_msats == 57000

tests/integration/test_admin_provider_balance.py

@@ -0,0 +1,80 @@
+from datetime import datetime, timedelta, timezone
+from unittest.mock import AsyncMock, patch
+
+import httpx
+import pytest
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.admin import admin_sessions
+from routstr.core.db import UpstreamProviderRow
+
+
+async def _create_routstr_provider() -> UpstreamProviderRow:
+    return UpstreamProviderRow(
+        provider_type="routstr",
+        base_url="https://upstream.example",
+        api_key="",
+        enabled=True,
+    )
+
+
+def _admin_headers() -> dict[str, str]:
+    token = "test-admin-token"
+    admin_sessions[token] = int(
+        (datetime.now(timezone.utc) + timedelta(minutes=5)).timestamp()
+    )
+    return {"Authorization": f"Bearer {token}"}
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_admin_routstr_provider_balance_timeout_returns_504(
+    integration_client: httpx.AsyncClient,
+    integration_session: AsyncSession,
+) -> None:
+    provider = await _create_routstr_provider()
+    integration_session.add(provider)
+    await integration_session.commit()
+    await integration_session.refresh(provider)
+
+    request = httpx.Request("GET", f"{provider.base_url}/v1/balance/info")
+    timeout_error = httpx.ConnectTimeout("Connect timeout", request=request)
+
+    with patch(
+        "httpx.AsyncHTTPTransport.handle_async_request",
+        new=AsyncMock(side_effect=timeout_error),
+    ):
+        response = await integration_client.get(
+            f"/admin/api/upstream-providers/{provider.id}/balance",
+            headers=_admin_headers(),
+        )
+
+    assert response.status_code == 504
+    assert response.json()["detail"] == "Timed out contacting upstream Routstr provider"
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_admin_routstr_provider_balance_request_error_returns_502(
+    integration_client: httpx.AsyncClient,
+    integration_session: AsyncSession,
+) -> None:
+    provider = await _create_routstr_provider()
+    integration_session.add(provider)
+    await integration_session.commit()
+    await integration_session.refresh(provider)
+
+    request = httpx.Request("GET", f"{provider.base_url}/v1/balance/info")
+    request_error = httpx.ConnectError("Connection failed", request=request)
+
+    with patch(
+        "httpx.AsyncHTTPTransport.handle_async_request",
+        new=AsyncMock(side_effect=request_error),
+    ):
+        response = await integration_client.get(
+            f"/admin/api/upstream-providers/{provider.id}/balance",
+            headers=_admin_headers(),
+        )
+
+    assert response.status_code == 502
+    assert response.json()["detail"] == "Failed to contact upstream Routstr provider"

tests/integration/test_balance_negative_on_cost_overrun.py

@@ -0,0 +1,369 @@
+"""
+Integration tests for the balance-goes-negative bug in adjust_payment_for_tokens.
+
+Root cause: when actual token cost exceeds the discounted reservation
+(cost_difference > 0, caused by tolerance_percentage discounting the reservation),
+the finalization UPDATE had no WHERE guard on balance, allowing balance to go negative.
+
+Fix: added `.where(col(ApiKey.balance) >= total_cost_msats)` so the UPDATE is a no-op
+when balance is insufficient, then falls back to charging only deducted_max_cost.
+"""
+
+import uuid
+from unittest.mock import patch
+
+import pytest
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.db import ApiKey
+from routstr.payment.cost_calculation import CostData
+
+
+def _make_key(balance: int, reserved: int) -> ApiKey:
+    return ApiKey(
+        hashed_key=f"test_{uuid.uuid4().hex}",
+        balance=balance,
+        reserved_balance=reserved,
+        total_spent=0,
+        total_requests=1,
+    )
+
+
+async def _refresh(session: AsyncSession, key: ApiKey) -> ApiKey:
+    await session.refresh(key)
+    return key
+
+
+# ---------------------------------------------------------------------------
+# Helper: build a CostData where token cost > deducted_max_cost
+# ---------------------------------------------------------------------------
+
+def _cost_data(total_msats: int) -> CostData:
+    return CostData(
+        base_msats=0,
+        input_msats=total_msats // 2,
+        output_msats=total_msats - total_msats // 2,
+        total_msats=total_msats,
+        total_usd=0.0,
+        input_tokens=100,
+        output_tokens=100,
+    )
+
+
+# ---------------------------------------------------------------------------
+# Test 1 — exact reproduction of the bug
+#
+# Setup: balance == deducted_max_cost (user has just enough for the reservation,
+#         nothing extra). Actual token cost is 1% higher (tolerance_percentage).
+#
+# Before fix: balance -= total_cost_msats → goes negative.
+# After fix:  WHERE balance >= total_cost_msats fails → fallback charges
+#             deducted_max_cost → balance reaches 0, never negative.
+# ---------------------------------------------------------------------------
+
+@pytest.mark.asyncio
+async def test_balance_never_negative_when_cost_exceeds_reservation(
+    integration_session: AsyncSession,
+) -> None:
+    """Balance must not go negative when actual token cost > discounted reservation."""
+    from routstr.auth import adjust_payment_for_tokens
+
+    deducted_max_cost = 990  # reserved (1% below true max of 1000)
+    actual_token_cost = 1000  # actual cost at true max
+
+    # User has balance exactly equal to the reservation — tight budget
+    key = _make_key(balance=deducted_max_cost, reserved=deducted_max_cost)
+    integration_session.add(key)
+    await integration_session.commit()
+
+    response_data = {"model": "test-model", "usage": {"prompt_tokens": 100, "completion_tokens": 100}}
+
+    with patch(
+        "routstr.auth.calculate_cost",
+        return_value=_cost_data(actual_token_cost),
+    ):
+        await adjust_payment_for_tokens(key, response_data, integration_session, deducted_max_cost)
+
+    await _refresh(integration_session, key)
+
+    assert key.balance >= 0, f"Balance went negative: {key.balance}"
+    assert key.reserved_balance >= 0, f"Reserved balance went negative: {key.reserved_balance}"
+    assert key.reserved_balance == 0, "Reservation must be fully released after finalization"
+
+
+# ---------------------------------------------------------------------------
+# Test 2 — balance is ZERO after the reservation is accounted for
+#           (absolute floor case)
+# ---------------------------------------------------------------------------
+
+@pytest.mark.asyncio
+async def test_balance_floor_at_zero_on_overrun(
+    integration_session: AsyncSession,
+) -> None:
+    """When balance exactly covers deducted_max_cost and cost overruns, balance reaches 0 not negative."""
+    from routstr.auth import adjust_payment_for_tokens
+
+    deducted_max_cost = 500
+    actual_token_cost = 550  # 10% overrun
+
+    key = _make_key(balance=500, reserved=500)
+    integration_session.add(key)
+    await integration_session.commit()
+
+    response_data = {"model": "test-model", "usage": {"prompt_tokens": 50, "completion_tokens": 50}}
+
+    with patch(
+        "routstr.auth.calculate_cost",
+        return_value=_cost_data(actual_token_cost),
+    ):
+        await adjust_payment_for_tokens(key, response_data, integration_session, deducted_max_cost)
+
+    await _refresh(integration_session, key)
+
+    assert key.balance == 0, (
+        f"Expected balance=0 (charged deducted_max_cost fallback), got {key.balance}"
+    )
+    assert key.reserved_balance == 0, f"Reserved balance should be 0, got {key.reserved_balance}"
+    # Fallback charges deducted_max_cost
+    assert key.total_spent == deducted_max_cost, (
+        f"Expected total_spent={deducted_max_cost}, got {key.total_spent}"
+    )
+
+
+# ---------------------------------------------------------------------------
+# Test 3 — balance has enough room: full token cost should be charged
+# ---------------------------------------------------------------------------
+
+@pytest.mark.asyncio
+async def test_full_cost_charged_when_balance_sufficient_for_overrun(
+    integration_session: AsyncSession,
+) -> None:
+    """When balance covers total_cost_msats, the full amount is charged (not just deducted_max_cost)."""
+    from routstr.auth import adjust_payment_for_tokens
+
+    deducted_max_cost = 990
+    actual_token_cost = 1000
+
+    # User has extra balance beyond the reservation
+    key = _make_key(balance=2000, reserved=990)
+    integration_session.add(key)
+    await integration_session.commit()
+
+    response_data = {"model": "test-model", "usage": {"prompt_tokens": 100, "completion_tokens": 100}}
+
+    with patch(
+        "routstr.auth.calculate_cost",
+        return_value=_cost_data(actual_token_cost),
+    ):
+        await adjust_payment_for_tokens(key, response_data, integration_session, deducted_max_cost)
+
+    await _refresh(integration_session, key)
+
+    assert key.balance >= 0, f"Balance went negative: {key.balance}"
+    assert key.reserved_balance == 0, f"Reservation not released: {key.reserved_balance}"
+    assert key.total_spent == actual_token_cost, (
+        f"Expected full charge of {actual_token_cost}, got {key.total_spent}"
+    )
+    assert key.balance == 2000 - actual_token_cost, (
+        f"Expected balance={2000 - actual_token_cost}, got {key.balance}"
+    )
+
+
+# ---------------------------------------------------------------------------
+# Test 4 — concurrent finalizations with cost overrun
+#
+# Multiple requests finish concurrently. Each has a small overrun.
+# None should drive balance negative.
+# ---------------------------------------------------------------------------
+
+@pytest.mark.asyncio
+async def test_concurrent_cost_overruns_never_negative(
+    integration_session: AsyncSession,
+    patched_db_engine: None,
+) -> None:
+    """Concurrent finalization with cost overruns must never produce negative balance."""
+    import asyncio
+
+    from routstr.auth import adjust_payment_for_tokens, pay_for_request
+    from routstr.core.db import create_session
+
+    deducted_max_cost = 990
+    actual_token_cost = 1000
+    n_requests = 5
+
+    # Fund the key with exactly enough for n_requests reservations + a tiny buffer
+    starting_balance = deducted_max_cost * n_requests
+    key_hash = f"test_concurrent_{uuid.uuid4().hex}"
+
+    async with create_session() as session:
+        key = ApiKey(
+            hashed_key=key_hash,
+            balance=starting_balance,
+            reserved_balance=0,
+            total_spent=0,
+            total_requests=0,
+        )
+        session.add(key)
+        await session.commit()
+
+    # Reserve n_requests slots (sequentially, as pay_for_request is atomic)
+    async with create_session() as session:
+        key_to_reserve = await session.get(ApiKey, key_hash)
+        assert key_to_reserve is not None
+        for _ in range(n_requests):
+            await pay_for_request(key_to_reserve, deducted_max_cost, session)
+            await session.refresh(key_to_reserve)
+
+    # Now finalize all concurrently with cost overrun
+    async def finalize() -> None:
+        response_data = {
+            "model": "test-model",
+            "usage": {"prompt_tokens": 100, "completion_tokens": 100},
+        }
+        async with create_session() as session:
+            fresh_key = await session.get(ApiKey, key_hash)
+            assert fresh_key is not None
+            with patch(
+                "routstr.auth.calculate_cost",
+                return_value=_cost_data(actual_token_cost),
+            ):
+                await adjust_payment_for_tokens(
+                    fresh_key, response_data, session, deducted_max_cost
+                )
+
+    await asyncio.gather(*[finalize() for _ in range(n_requests)])
+
+    async with create_session() as session:
+        final_key = await session.get(ApiKey, key_hash)
+        assert final_key is not None
+
+    assert final_key.balance >= 0, (
+        f"Balance went negative after concurrent overruns: {final_key.balance}"
+    )
+    assert final_key.reserved_balance == 0, (
+        f"Reserved balance not fully released: {final_key.reserved_balance}"
+    )
+    assert final_key.total_spent <= starting_balance, (
+        f"Total spent ({final_key.total_spent}) exceeds starting balance ({starting_balance})"
+    )
+    # Every request must have been charged at least deducted_max_cost — no free inference.
+    assert final_key.total_spent == starting_balance, (
+        f"Expected total_spent={starting_balance} (all {n_requests} reservations charged), "
+        f"got {final_key.total_spent} — at least one request got free inference"
+    )
+
+
+# ---------------------------------------------------------------------------
+# Test 5 — overrun with no balance at all (reserved_balance == balance)
+#           simulates a user who topped up to exactly the reservation floor
+# ---------------------------------------------------------------------------
+
+@pytest.mark.asyncio
+async def test_zero_free_balance_overrun_is_safe(
+    integration_session: AsyncSession,
+) -> None:
+    """User with zero free balance (all reserved) should never go negative on overrun."""
+    from routstr.auth import adjust_payment_for_tokens
+
+    deducted_max_cost = 1000
+    actual_token_cost = 1050
+
+    # balance == reserved_balance: zero free balance
+    key = _make_key(balance=1000, reserved=1000)
+    integration_session.add(key)
+    await integration_session.commit()
+
+    response_data = {"model": "test-model", "usage": {"prompt_tokens": 50, "completion_tokens": 100}}
+
+    with patch(
+        "routstr.auth.calculate_cost",
+        return_value=_cost_data(actual_token_cost),
+    ):
+        await adjust_payment_for_tokens(key, response_data, integration_session, deducted_max_cost)
+
+    await _refresh(integration_session, key)
+
+    assert key.balance >= 0, f"Balance went negative: {key.balance}"
+    assert key.reserved_balance >= 0, f"Reserved balance went negative: {key.reserved_balance}"
+
+
+# ---------------------------------------------------------------------------
+# Test 6 — parallel requests: second finalization must not get free inference
+#
+# Root cause of the bug fixed in auth.py:
+#   `.where(col(ApiKey.balance) >= total_cost_msats)` ignores other requests'
+#   reservations, so after Request A charges total_cost_msats, balance can drop
+#   below deducted_max_cost, causing Request B's fallback to release for free.
+#
+# Fix: use `balance - reserved_balance + deducted_max_cost >= total_cost_msats`
+#   so the check accounts for concurrent reservations.
+# ---------------------------------------------------------------------------
+
+
+@pytest.mark.asyncio
+async def test_parallel_requests_no_free_inference(
+    integration_session: AsyncSession,
+    patched_db_engine: None,
+) -> None:
+    """Second parallel finalization must be charged even when first depleted free balance."""
+    import asyncio
+
+    from routstr.auth import adjust_payment_for_tokens
+    from routstr.core.db import create_session
+
+    deducted_max_cost = 100
+    actual_token_cost = 150  # overrun: 50 more than reserved
+
+    # Fund the key with exactly 2 * deducted_max_cost.
+    # Both requests pre-reserved 100 each → balance=200, reserved=200, free=0.
+    # Old check (balance >= total_cost_msats):
+    #   Request A: 200 >= 150 ✓ → charges 150 → balance=50, reserved=100
+    #   Request B: 50 >= 150 ✗ → fallback: 50 >= 100 ✗ → releases FREE
+    # New check (balance - reserved + deducted >= total_cost_msats):
+    #   Both fall to fallback (0 free balance).
+    #   Both charge deducted_max_cost=100 → total_spent=200, balance=0.
+    starting_balance = deducted_max_cost * 2
+    key_hash = f"test_parallel_no_free_{uuid.uuid4().hex}"
+
+    async with create_session() as session:
+        key = ApiKey(
+            hashed_key=key_hash,
+            balance=starting_balance,
+            reserved_balance=deducted_max_cost * 2,  # both slots pre-reserved
+            total_spent=0,
+            total_requests=2,
+        )
+        session.add(key)
+        await session.commit()
+
+    async def finalize() -> None:
+        response_data = {
+            "model": "test-model",
+            "usage": {"prompt_tokens": 50, "completion_tokens": 100},
+        }
+        async with create_session() as session:
+            fresh_key = await session.get(ApiKey, key_hash)
+            assert fresh_key is not None
+            with patch(
+                "routstr.auth.calculate_cost",
+                return_value=_cost_data(actual_token_cost),
+            ):
+                await adjust_payment_for_tokens(
+                    fresh_key, response_data, session, deducted_max_cost
+                )
+
+    await asyncio.gather(finalize(), finalize())
+
+    async with create_session() as session:
+        final_key = await session.get(ApiKey, key_hash)
+        assert final_key is not None
+
+    assert final_key.balance >= 0, f"Balance went negative: {final_key.balance}"
+    assert final_key.reserved_balance == 0, (
+        f"Reserved balance not released: {final_key.reserved_balance}"
+    )
+    # Both requests must have been charged — no free inference.
+    assert final_key.total_spent == starting_balance, (
+        f"Expected total_spent={starting_balance} (both reservations charged), "
+        f"got {final_key.total_spent} — one request got free inference"
+    )

tests/integration/test_child_keys.py

@@ -1,3 +1,4 @@
+import asyncio
 import secrets
 from typing import Any
 
@@ -7,7 +8,7 @@ from sqlmodel.ext.asyncio.session import AsyncSession
 
 from routstr.auth import adjust_payment_for_tokens, pay_for_request
 from routstr.balance import ChildKeyRequest, create_child_key
-from routstr.core.db import ApiKey
+from routstr.core.db import ApiKey, create_session
 from routstr.core.settings import settings
 
 
@@ -119,6 +120,54 @@ async def test_child_key_insufficient_balance(
     assert exc.value.status_code == 402
 
 
+@pytest.mark.asyncio
+async def test_concurrent_child_key_creation_is_atomic(
+    patched_db_engine: None,
+) -> None:
+    """Two concurrent create_child_key() calls with balance for exactly one must
+    result in exactly one success and one 402, with the parent balance deducted
+    only once."""
+    child_key_cost = 1000
+    settings.child_key_cost = child_key_cost
+
+    parent_hash = f"parent_concurrent_{secrets.token_hex(8)}"
+    async with create_session() as session:
+        parent = ApiKey(hashed_key=parent_hash, balance=child_key_cost)
+        session.add(parent)
+        await session.commit()
+
+    results: list[str] = []
+
+    async def attempt() -> None:
+        async with create_session() as session:
+            fresh_parent = await session.get(ApiKey, parent_hash)
+            assert fresh_parent is not None
+            try:
+                await create_child_key(ChildKeyRequest(count=1), fresh_parent, session)
+                results.append("success")
+            except HTTPException as exc:
+                assert exc.status_code == 402
+                results.append("blocked")
+
+    await asyncio.gather(attempt(), attempt())
+
+    assert sorted(results) == ["blocked", "success"], (
+        f"Expected exactly one success and one 402, got: {results}"
+    )
+
+    async with create_session() as session:
+        final = await session.get(ApiKey, parent_hash)
+        assert final is not None
+
+    assert final.balance == 0, (
+        f"Balance should be fully deducted once: expected 0, got {final.balance}"
+    )
+    assert final.total_spent == child_key_cost, (
+        f"total_spent should equal one deduction: expected {child_key_cost}, "
+        f"got {final.total_spent}"
+    )
+
+
 @pytest.mark.asyncio
 async def test_child_key_cannot_create_child(integration_session: AsyncSession) -> None:
     parent_key = ApiKey(

tests/integration/test_child_keys_api.py

@@ -69,9 +69,14 @@ async def test_wallet_info_child_key_no_child_keys(
     info_response = await integration_client.get("/v1/wallet/info")
     assert info_response.status_code == 200
     info_data = info_response.json()
+    parent_key = authenticated_client._test_api_key  # type: ignore[attr-defined]
+    parent_key_hash = parent_key.removeprefix("sk-")
 
     assert info_data["is_child"] is True
     assert "child_keys" not in info_data
+    assert "parent_key" not in info_data
+    assert info_data["parent_key_preview"] == parent_key_hash[:8] + "..."
+    assert info_data["parent_key_preview"] not in {parent_key, parent_key_hash}
 
 
 @pytest.mark.integration

tests/integration/test_cli_tokens.py

@@ -0,0 +1,302 @@
+"""Integration tests for CLI token management (/admin/api/cli-tokens).
+
+Covers:
+- GET    /admin/api/cli-tokens           — list (preview only, no full token)
+- POST   /admin/api/cli-tokens           — create (returns full token once)
+- DELETE /admin/api/cli-tokens/{id}      — revoke
+- Using a CLI token as Bearer auth against admin endpoints
+- Expiry enforcement (expired tokens are rejected by require_admin_api)
+- last_used_at bump on successful use
+- Auth failures: missing token, wrong token, revoked token
+"""
+
+from __future__ import annotations
+
+import secrets
+import time
+from typing import AsyncGenerator
+
+import pytest
+import pytest_asyncio
+from httpx import AsyncClient
+from sqlmodel import select
+
+from routstr.core.admin import admin_sessions
+from routstr.core.db import AsyncSession, CliToken
+
+# ──────────────────────────────────────────────────────────────────────────────
+# Fixtures
+# ──────────────────────────────────────────────────────────────────────────────
+
+@pytest_asyncio.fixture
+async def admin_session_token() -> AsyncGenerator[str, None]:
+    """Inject a short-lived admin session token into admin_sessions."""
+    token = secrets.token_urlsafe(24)
+    admin_sessions[token] = int(time.time()) + 3600
+    yield token
+    admin_sessions.pop(token, None)
+
+
+@pytest_asyncio.fixture
+async def admin_client(
+    integration_client: AsyncClient, admin_session_token: str
+) -> AsyncClient:
+    """An integration_client pre-authenticated with an admin session token."""
+    integration_client.headers["Authorization"] = f"Bearer {admin_session_token}"
+    return integration_client
+
+
+# ──────────────────────────────────────────────────────────────────────────────
+# Creation
+# ──────────────────────────────────────────────────────────────────────────────
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_create_cli_token_returns_full_token_once(
+    admin_client: AsyncClient,
+) -> None:
+    """POST /admin/api/cli-tokens returns the raw token only on creation."""
+    resp = await admin_client.post(
+        "/admin/api/cli-tokens",
+        json={"name": "my-laptop"},
+    )
+    assert resp.status_code == 200
+    body = resp.json()
+
+    assert body["name"] == "my-laptop"
+    assert isinstance(body["id"], str) and body["id"]
+    assert isinstance(body["token"], str) and len(body["token"]) >= 32
+    assert body["expires_at"] is None
+    assert isinstance(body["created_at"], int)
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_create_cli_token_with_expiry(admin_client: AsyncClient) -> None:
+    """expires_in_days sets expires_at ~= now + days * 86400."""
+    before = int(time.time())
+    resp = await admin_client.post(
+        "/admin/api/cli-tokens",
+        json={"name": "ci-runner", "expires_in_days": 7},
+    )
+    assert resp.status_code == 200
+    body = resp.json()
+
+    assert body["expires_at"] is not None
+    delta = body["expires_at"] - before
+    # Allow 10s jitter around 7 * 86400
+    assert 7 * 86400 - 10 <= delta <= 7 * 86400 + 10
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_create_cli_token_rejects_empty_name(
+    admin_client: AsyncClient,
+) -> None:
+    resp = await admin_client.post(
+        "/admin/api/cli-tokens", json={"name": "   "}
+    )
+    assert resp.status_code == 400
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_create_cli_token_requires_admin(
+    integration_client: AsyncClient,
+) -> None:
+    """No admin token / no bearer → 403."""
+    resp = await integration_client.post(
+        "/admin/api/cli-tokens", json={"name": "no-auth"}
+    )
+    assert resp.status_code == 403
+
+
+# ──────────────────────────────────────────────────────────────────────────────
+# Listing
+# ──────────────────────────────────────────────────────────────────────────────
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_list_cli_tokens_returns_preview_not_full_token(
+    admin_client: AsyncClient,
+) -> None:
+    """Listing never leaks the raw token."""
+    create = await admin_client.post(
+        "/admin/api/cli-tokens", json={"name": "secret-keeper"}
+    )
+    assert create.status_code == 200
+    full_token = create.json()["token"]
+
+    resp = await admin_client.get("/admin/api/cli-tokens")
+    assert resp.status_code == 200
+    items = resp.json()
+    assert any(t["name"] == "secret-keeper" for t in items)
+
+    for t in items:
+        # No 'token' field, only 'token_preview'
+        assert "token" not in t
+        assert "token_preview" in t
+        assert full_token not in t["token_preview"]
+        assert "..." in t["token_preview"]
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_list_cli_tokens_requires_admin(
+    integration_client: AsyncClient,
+) -> None:
+    resp = await integration_client.get("/admin/api/cli-tokens")
+    assert resp.status_code == 403
+
+
+# ──────────────────────────────────────────────────────────────────────────────
+# Using a CLI token as admin auth
+# ──────────────────────────────────────────────────────────────────────────────
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_cli_token_authorizes_admin_endpoints(
+    admin_client: AsyncClient,
+    integration_client: AsyncClient,
+    integration_session: AsyncSession,
+) -> None:
+    """A freshly-created CLI token can be used as Bearer on admin endpoints."""
+    create = await admin_client.post(
+        "/admin/api/cli-tokens", json={"name": "cli-auth"}
+    )
+    assert create.status_code == 200
+    cli_token = create.json()["token"]
+    token_id = create.json()["id"]
+
+    # Use a NEW client to isolate the header from admin_session_token
+    integration_client.headers["Authorization"] = f"Bearer {cli_token}"
+    resp = await integration_client.get("/admin/api/cli-tokens")
+    assert resp.status_code == 200
+
+    # last_used_at should be populated after use
+    row = await integration_session.get(CliToken, token_id)
+    assert row is not None
+    assert row.last_used_at is not None
+    assert row.last_used_at >= row.created_at
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_expired_cli_token_is_rejected(
+    admin_client: AsyncClient,
+    integration_client: AsyncClient,
+    integration_session: AsyncSession,
+) -> None:
+    """A CLI token with expires_at in the past → 403."""
+    create = await admin_client.post(
+        "/admin/api/cli-tokens",
+        json={"name": "will-expire", "expires_in_days": 1},
+    )
+    assert create.status_code == 200
+    cli_token = create.json()["token"]
+    token_id = create.json()["id"]
+
+    # Force-expire it in the DB
+    row = await integration_session.get(CliToken, token_id)
+    assert row is not None
+    row.expires_at = int(time.time()) - 1
+    integration_session.add(row)
+    await integration_session.commit()
+
+    integration_client.headers["Authorization"] = f"Bearer {cli_token}"
+    resp = await integration_client.get("/admin/api/cli-tokens")
+    assert resp.status_code == 403
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_invalid_bearer_token_is_rejected(
+    integration_client: AsyncClient,
+) -> None:
+    integration_client.headers["Authorization"] = "Bearer not-a-real-token"
+    resp = await integration_client.get("/admin/api/cli-tokens")
+    assert resp.status_code == 403
+
+
+# ──────────────────────────────────────────────────────────────────────────────
+# Revocation
+# ──────────────────────────────────────────────────────────────────────────────
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_revoke_cli_token_removes_auth(
+    admin_client: AsyncClient,
+    integration_client: AsyncClient,
+    integration_session: AsyncSession,
+) -> None:
+    """After DELETE, the token no longer authorizes."""
+    create = await admin_client.post(
+        "/admin/api/cli-tokens", json={"name": "to-revoke"}
+    )
+    token_id = create.json()["id"]
+    cli_token = create.json()["token"]
+
+    revoke = await admin_client.delete(f"/admin/api/cli-tokens/{token_id}")
+    assert revoke.status_code == 200
+    assert revoke.json() == {"ok": True, "deleted_id": token_id}
+
+    # Row is gone
+    row = await integration_session.get(CliToken, token_id)
+    assert row is None
+
+    # Can no longer be used for auth
+    integration_client.headers["Authorization"] = f"Bearer {cli_token}"
+    resp = await integration_client.get("/admin/api/cli-tokens")
+    assert resp.status_code == 403
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_revoke_unknown_cli_token_returns_404(
+    admin_client: AsyncClient,
+) -> None:
+    resp = await admin_client.delete("/admin/api/cli-tokens/does-not-exist")
+    assert resp.status_code == 404
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_revoke_cli_token_requires_admin(
+    integration_client: AsyncClient,
+) -> None:
+    resp = await integration_client.delete("/admin/api/cli-tokens/anything")
+    assert resp.status_code == 403
+
+
+# ──────────────────────────────────────────────────────────────────────────────
+# Lifecycle / uniqueness
+# ──────────────────────────────────────────────────────────────────────────────
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_multiple_tokens_are_independent(
+    admin_client: AsyncClient,
+    integration_session: AsyncSession,
+) -> None:
+    """Creating N tokens yields N unique tokens that all live in DB."""
+    names = ["dev-a", "dev-b", "dev-c"]
+    raw_tokens: list[str] = []
+    ids: list[str] = []
+    for name in names:
+        r = await admin_client.post(
+            "/admin/api/cli-tokens", json={"name": name}
+        )
+        assert r.status_code == 200
+        raw_tokens.append(r.json()["token"])
+        ids.append(r.json()["id"])
+
+    # All unique
+    assert len(set(raw_tokens)) == len(raw_tokens)
+    assert len(set(ids)) == len(ids)
+
+    # All in DB
+    result = await integration_session.exec(
+        select(CliToken).where(CliToken.name.in_(names))  # type: ignore[attr-defined]
+    )
+    rows = result.all()
+    assert {r.name for r in rows} == set(names)

tests/integration/test_free_response_stale_reservation.py

@@ -0,0 +1,146 @@
+"""Regression tests for charging after stale reservation cleanup."""
+
+import time
+import uuid
+from unittest.mock import patch
+
+import pytest
+from sqlmodel.ext.asyncio.session import AsyncSession
+
+from routstr.core.db import ApiKey
+from routstr.payment.cost_calculation import CostData
+
+
+def _make_key(balance: int, reserved: int) -> ApiKey:
+    return ApiKey(
+        hashed_key=f"test_{uuid.uuid4().hex}",
+        balance=balance,
+        reserved_balance=reserved,
+        total_spent=0,
+        total_requests=1,
+    )
+
+
+def _cost_data(total_msats: int) -> CostData:
+    return CostData(
+        base_msats=0,
+        input_msats=total_msats // 2,
+        output_msats=total_msats - total_msats // 2,
+        total_msats=total_msats,
+        total_usd=0.0,
+        input_tokens=100,
+        output_tokens=100,
+    )
+
+
+@pytest.mark.asyncio
+async def test_overrun_charges_after_reservation_swept(
+    integration_session: AsyncSession,
+) -> None:
+    """Overrun finalize must charge even when the reservation was already released."""
+    from routstr.auth import adjust_payment_for_tokens
+
+    deducted_max_cost = 990  # discounted reservation
+    actual_token_cost = 1000  # actual cost overruns the reservation
+
+    # Sweeper has zeroed reserved_balance but left balance untouched.
+    key = _make_key(balance=1000, reserved=0)
+    integration_session.add(key)
+    await integration_session.commit()
+
+    response_data = {
+        "model": "test-model",
+        "usage": {"prompt_tokens": 100, "completion_tokens": 100},
+    }
+
+    with patch(
+        "routstr.auth.calculate_cost",
+        return_value=_cost_data(actual_token_cost),
+    ):
+        await adjust_payment_for_tokens(
+            key, response_data, integration_session, deducted_max_cost
+        )
+
+    await integration_session.refresh(key)
+
+    assert key.total_spent == actual_token_cost, (
+        f"Request was not billed (total_spent={key.total_spent}) — free response bug"
+    )
+    assert key.balance == 1000 - actual_token_cost, (
+        f"Balance not charged: {key.balance}"
+    )
+    assert key.balance >= 0
+    assert key.reserved_balance == 0
+
+
+@pytest.mark.asyncio
+async def test_free_response_path_closed_end_to_end(
+    integration_session: AsyncSession,
+    patched_db_engine: None,
+) -> None:
+    """A reservation released by the real sweeper must not yield a free response."""
+    from routstr.auth import adjust_payment_for_tokens, pay_for_request
+    from routstr.core.db import create_session, release_stale_reservations
+
+    deducted_max_cost = 990
+    actual_token_cost = 1000
+    key_hash = f"test_sweep_{uuid.uuid4().hex}"
+
+    async with create_session() as session:
+        session.add(
+            ApiKey(
+                hashed_key=key_hash,
+                balance=1000,
+                reserved_balance=0,
+                total_spent=0,
+                total_requests=0,
+            )
+        )
+        await session.commit()
+
+    # Reserve the request, then backdate reserved_at so the sweeper treats it as
+    # stale (simulates a stream that outlived stale_reservation_timeout_seconds).
+    async with create_session() as session:
+        key = await session.get(ApiKey, key_hash)
+        assert key is not None
+        await pay_for_request(key, deducted_max_cost, session)
+        await session.refresh(key)
+        assert key.reserved_balance == deducted_max_cost
+        key.reserved_at = int(time.time()) - 10_000
+        session.add(key)
+        await session.commit()
+
+    # Sweeper releases the stale reservation without charging.
+    async with create_session() as session:
+        released = await release_stale_reservations(session, max_age_seconds=300)
+        assert released == 1
+
+    async with create_session() as session:
+        key = await session.get(ApiKey, key_hash)
+        assert key is not None
+        assert key.reserved_balance == 0, "Precondition: sweeper zeroed the reservation"
+
+        response_data = {
+            "model": "test-model",
+            "usage": {"prompt_tokens": 100, "completion_tokens": 100},
+        }
+        with patch(
+            "routstr.auth.calculate_cost",
+            return_value=_cost_data(actual_token_cost),
+        ):
+            await adjust_payment_for_tokens(
+                key, response_data, session, deducted_max_cost
+            )
+
+    async with create_session() as session:
+        final = await session.get(ApiKey, key_hash)
+        assert final is not None
+
+    assert final.total_spent == actual_token_cost, (
+        f"Free response: total_spent={final.total_spent}, expected {actual_token_cost}"
+    )
+    assert final.balance == 1000 - actual_token_cost, (
+        f"Balance not charged after sweep: {final.balance}"
+    )
+    assert final.balance >= 0
+    assert final.reserved_balance == 0