fix alias not displaying correctly

# Conflicts:
#	routstr/upstream/base.py

migrations/versions/b9667ffc5701_alias_ids.py

@@ -0,0 +1,37 @@
+"""alias-ids
+
+Revision ID: b9667ffc5701
+Revises: lightning_invoices
+Create Date: 2025-12-25 19:30:44.673350
+"""
+
+import sqlalchemy as sa
+import sqlmodel
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "b9667ffc5701"
+down_revision = "lightning_invoices"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic ###
+    op.add_column(
+        "models",
+        sa.Column("canonical_slug", sqlmodel.sql.sqltypes.AutoString(), nullable=True),
+    )
+    op.add_column(
+        "models",
+        sa.Column("alias_ids", sqlmodel.sql.sqltypes.AutoString(), nullable=True),
+    )
+
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_column("models", "alias_ids")
+    op.drop_column("models", "canonical_slug")
+    # ### end Alembic commands ###

routstr/algorithm.py

@@ -206,19 +206,20 @@ def create_model_mappings(
         alias: str, model: "Model", provider: "BaseUpstreamProvider"
     ) -> None:
         """Set alias to model/provider if not set or if new model is preferred."""
-        existing_model = model_instances.get(alias)
+        alias_lower = alias.lower()
+        existing_model = model_instances.get(alias_lower)
         if not existing_model:
             # No existing mapping, set it
-            model_instances[alias] = model
-            provider_map[alias] = provider
+            model_instances[alias_lower] = model
+            provider_map[alias_lower] = provider
         else:
             # Check if candidate should replace existing
-            existing_provider = provider_map[alias]
+            existing_provider = provider_map[alias_lower]
             if should_prefer_model(
                 model, provider, existing_model, existing_provider, alias
             ):
-                model_instances[alias] = model
-                provider_map[alias] = provider
+                model_instances[alias_lower] = model
+                provider_map[alias_lower] = provider
 
     def process_provider_models(
         upstream: "BaseUpstreamProvider", is_openrouter: bool = False

routstr/core/admin.py

@@ -1493,20 +1493,8 @@ class ModelCreate(BaseModel):
     per_request_limits: dict[str, object] | None = None
     top_provider: dict[str, object] | None = None
     upstream_provider_id: int | None = None
-    enabled: bool = True
-
-
-class ModelUpdate(BaseModel):
-    id: str
-    name: str
-    description: str
-    created: int
-    context_length: int
-    architecture: dict[str, object]
-    pricing: dict[str, object]
-    per_request_limits: dict[str, object] | None = None
-    top_provider: dict[str, object] | None = None
-    upstream_provider_id: int | None = None
+    canonical_slug: str | None = None
+    alias_ids: list[str] | None = None
     enabled: bool = True
 
 
@@ -2416,44 +2404,80 @@ async def admin_upstream_providers(request: Request) -> str:
     "/api/upstream-providers/{provider_id}/models",
     dependencies=[Depends(require_admin_api)],
 )
-async def create_provider_model(
+async def upsert_provider_model(
     provider_id: int, payload: ModelCreate
 ) -> dict[str, object]:
+    print(payload)
+    logger.info(
+        f"UPSERT_PROVIDER_MODEL called: provider_id={provider_id}, model_id={payload.id}"
+    )
     async with create_session() as session:
         provider = await session.get(UpstreamProviderRow, provider_id)
         if not provider:
             raise HTTPException(status_code=404, detail="Provider not found")
 
-        exists = await session.get(ModelRow, (payload.id, provider_id))
-        if exists:
-            raise HTTPException(
-                status_code=409,
-                detail="Model with this ID already exists for this provider",
-            )
+        # Try to get existing model
+        existing_row = await session.get(ModelRow, (payload.id, provider_id))
 
-        row = ModelRow(
-            id=payload.id,
-            name=payload.name,
-            description=payload.description,
-            created=int(payload.created),
-            context_length=int(payload.context_length),
-            architecture=json.dumps(payload.architecture),
-            pricing=json.dumps(payload.pricing),
-            sats_pricing=None,
-            per_request_limits=(
+        if existing_row:
+            # Update existing model
+            logger.info(f"Updating existing model: {payload.id}")
+            existing_row.name = payload.name
+            existing_row.description = payload.description
+            existing_row.created = int(payload.created)
+            existing_row.context_length = int(payload.context_length)
+            existing_row.architecture = json.dumps(payload.architecture)
+            existing_row.pricing = json.dumps(payload.pricing)
+            existing_row.sats_pricing = None
+            existing_row.per_request_limits = (
                 json.dumps(payload.per_request_limits)
                 if payload.per_request_limits is not None
                 else None
-            ),
-            top_provider=(
+            )
+            existing_row.top_provider = (
                 json.dumps(payload.top_provider) if payload.top_provider else None
-            ),
-            upstream_provider_id=provider_id,
-            enabled=payload.enabled,
-        )
-        session.add(row)
-        await session.commit()
-        await session.refresh(row)
+            )
+            existing_row.canonical_slug = payload.canonical_slug
+            existing_row.alias_ids = (
+                json.dumps(payload.alias_ids) if payload.alias_ids else None
+            )
+            existing_row.enabled = payload.enabled
+
+            session.add(existing_row)
+            await session.commit()
+            await session.refresh(existing_row)
+            row = existing_row
+
+        else:
+            # Create new model
+            logger.info(f"Creating new model: {payload.id}")
+            row = ModelRow(
+                id=payload.id,
+                name=payload.name,
+                description=payload.description,
+                created=int(payload.created),
+                context_length=int(payload.context_length),
+                architecture=json.dumps(payload.architecture),
+                pricing=json.dumps(payload.pricing),
+                sats_pricing=None,
+                per_request_limits=(
+                    json.dumps(payload.per_request_limits)
+                    if payload.per_request_limits is not None
+                    else None
+                ),
+                top_provider=(
+                    json.dumps(payload.top_provider) if payload.top_provider else None
+                ),
+                canonical_slug=payload.canonical_slug,
+                alias_ids=(
+                    json.dumps(payload.alias_ids) if payload.alias_ids else None
+                ),
+                upstream_provider_id=provider_id,
+                enabled=payload.enabled,
+            )
+            session.add(row)
+            await session.commit()
+            await session.refresh(row)
 
     await refresh_model_maps()
     return _row_to_model(
@@ -2461,6 +2485,20 @@ async def create_provider_model(
     ).dict()  # type: ignore
 
 
+@admin_router.patch(
+    "/api/upstream-providers/{provider_id}/models/{model_id:path}",
+    dependencies=[Depends(require_admin_api)],
+)
+async def update_provider_model_legacy(
+    provider_id: int, model_id: str, payload: ModelCreate
+) -> dict[str, object]:
+    """Legacy PATCH endpoint - redirects to upsert POST endpoint for backward compatibility."""
+    logger.info(
+        f"LEGACY_PATCH_UPDATE called: provider_id={provider_id}, model_id={model_id}"
+    )
+    return await upsert_provider_model(provider_id, payload)
+
+
 @admin_router.get(
     "/api/upstream-providers/{provider_id}/models/{model_id:path}",
     dependencies=[Depends(require_admin_api)],
@@ -2481,76 +2519,6 @@ async def get_provider_model(provider_id: int, model_id: str) -> dict[str, objec
         ).dict()  # type: ignore
 
 
-@admin_router.patch(
-    "/api/upstream-providers/{provider_id}/models/{model_id:path}",
-    dependencies=[Depends(require_admin_api)],
-)
-async def update_provider_model(
-    provider_id: int, model_id: str, payload: ModelUpdate
-) -> dict[str, object]:
-    if payload.id != model_id:
-        raise HTTPException(status_code=400, detail="Path id does not match payload id")
-
-    async with create_session() as session:
-        provider = await session.get(UpstreamProviderRow, provider_id)
-        if not provider:
-            raise HTTPException(status_code=404, detail="Provider not found")
-
-        row = await session.get(ModelRow, (model_id, provider_id))
-        if not row:
-            raise HTTPException(
-                status_code=404, detail="Model not found for this provider"
-            )
-
-        row.name = payload.name
-        row.description = payload.description
-        row.created = int(payload.created)
-        row.context_length = int(payload.context_length)
-        row.architecture = json.dumps(payload.architecture)
-        row.pricing = json.dumps(payload.pricing)
-        row.sats_pricing = None
-        row.per_request_limits = (
-            json.dumps(payload.per_request_limits)
-            if payload.per_request_limits is not None
-            else None
-        )
-        row.top_provider = (
-            json.dumps(payload.top_provider) if payload.top_provider else None
-        )
-        was_disabled = not row.enabled
-        row.enabled = payload.enabled
-
-        session.add(row)
-        await session.commit()
-        await session.refresh(row)
-
-    if was_disabled and payload.enabled:
-        from ..payment.models import _cleanup_enabled_models_once
-
-        try:
-            await _cleanup_enabled_models_once()
-        except Exception as e:
-            logger.warning(
-                f"Failed to run model cleanup after enabling: {e}",
-                extra={"model_id": model_id, "error": str(e)},
-            )
-
-    await refresh_model_maps()
-    return _row_to_model(
-        row, apply_provider_fee=True, provider_fee=provider.provider_fee
-    ).dict()  # type: ignore
-
-
-@admin_router.put(
-    "/api/upstream-providers/{provider_id}/models/{model_id:path}",
-    dependencies=[Depends(require_admin_api)],
-)
-async def update_provider_model_put(
-    provider_id: int, model_id: str, payload: ModelUpdate
-) -> dict[str, object]:
-    return await update_provider_model(provider_id, model_id, payload)
-
-
 @admin_router.delete(
     "/api/upstream-providers/{provider_id}/models/{model_id:path}",
     dependencies=[Depends(require_admin_api)],

routstr/core/db.py

@@ -68,6 +68,10 @@ class ModelRow(SQLModel, table=True):  # type: ignore
     sats_pricing: str | None = Field(default=None)
     per_request_limits: str | None = Field(default=None)
     top_provider: str | None = Field(default=None)
+    canonical_slug: str | None = Field(default=None, description="Canonical model slug")
+    alias_ids: str | None = Field(
+        default=None, description="JSON array of model alias IDs"
+    )
     enabled: bool = Field(default=True, description="Whether this model is enabled")
     upstream_provider: "UpstreamProviderRow" = Relationship(back_populates="models")
 

routstr/core/main.py

@@ -14,7 +14,6 @@ from ..balance import balance_router, deprecated_wallet_router
 from ..discovery import providers_cache_refresher, providers_router
 from ..nip91 import announce_provider
 from ..payment.models import (
-    cleanup_enabled_models_periodically,
     models_router,
     update_sats_pricing,
 )
@@ -49,7 +48,6 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
     nip91_task = None
     providers_task = None
     models_refresh_task = None
-    models_cleanup_task = None
     model_maps_refresh_task = None
 
     try:
@@ -86,7 +84,6 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
             models_refresh_task = asyncio.create_task(
                 refresh_upstreams_models_periodically(get_upstreams())
             )
-        models_cleanup_task = asyncio.create_task(cleanup_enabled_models_periodically())
         model_maps_refresh_task = asyncio.create_task(refresh_model_maps_periodically())
         payout_task = asyncio.create_task(periodic_payout())
         if global_settings.nsec:
@@ -125,8 +122,6 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
             providers_task.cancel()
         if models_refresh_task is not None:
             models_refresh_task.cancel()
-        if models_cleanup_task is not None:
-            models_cleanup_task.cancel()
         if model_maps_refresh_task is not None:
             model_maps_refresh_task.cancel()
 
@@ -144,8 +139,6 @@ async def lifespan(_: FastAPI) -> AsyncGenerator[None, None]:
                 tasks_to_wait.append(providers_task)
             if models_refresh_task is not None:
                 tasks_to_wait.append(models_refresh_task)
-            if models_cleanup_task is not None:
-                tasks_to_wait.append(models_cleanup_task)
             if model_maps_refresh_task is not None:
                 tasks_to_wait.append(model_maps_refresh_task)
 

routstr/payment/cost_calculation.py

@@ -5,6 +5,7 @@ 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
 
 logger = get_logger(__name__)
 
@@ -64,6 +65,56 @@ async def calculate_cost(  # todo: can be sync
         )
         return cost_data
 
+    usage_data = response_data["usage"]
+
+    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
+
+    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",
+                extra={
+                    "usd_cost": usd_cost,
+                    "cost_in_sats": cost_in_sats,
+                    "cost_in_msats": cost_in_msats,
+                    "model": response_data.get("model", "unknown"),
+                },
+            )
+
+            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,
+            )
+        except Exception as e:
+            logger.warning(
+                "Error calculating cost from usage data",
+                extra={
+                    "error": str(e),
+                    "usd_cost": usd_cost,
+                    "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
     )
@@ -129,10 +180,19 @@ async def calculate_cost(  # todo: can be sync
         )
         return cost_data
 
-    input_tokens = response_data.get("usage", {}).get("prompt_tokens", 0)
-    output_tokens = response_data.get("usage", {}).get("completion_tokens", 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)
+    )
 
     input_msats = round(input_tokens / 1000 * MSATS_PER_1K_INPUT_TOKENS, 3)
+
     output_msats = round(output_tokens / 1000 * MSATS_PER_1K_OUTPUT_TOKENS, 3)
     token_based_cost = math.ceil(input_msats + output_msats)
 

routstr/payment/models.py

@@ -5,10 +5,9 @@ import random
 import httpx
 from fastapi import APIRouter, Depends
 from pydantic.v1 import BaseModel
-from sqlmodel import select
 from sqlmodel.ext.asyncio.session import AsyncSession
 
-from ..core.db import ModelRow, create_session, get_session
+from ..core.db import ModelRow, get_session
 from ..core.logging import get_logger
 from ..core.settings import settings
 from .price import sats_usd_price
@@ -93,12 +92,32 @@ async def async_fetch_openrouter_models(source_filter: str | None = None) -> lis
 
     try:
         async with httpx.AsyncClient() as client:
-            response = await client.get(f"{base_url}/models", timeout=30)
-            response.raise_for_status()
-            data = response.json()
+            models_response, embeddings_response = await asyncio.gather(
+                client.get(f"{base_url}/models", timeout=30),
+                client.get(f"{base_url}/embeddings/models", timeout=30),
+                return_exceptions=True,
+            )
+
+            def process_models_response(
+                response: httpx.Response | BaseException,
+            ) -> list[dict]:
+                if not isinstance(response, BaseException):
+                    response.raise_for_status()
+                    data = response.json()
+                    return [
+                        model
+                        for model in data.get("data", [])
+                        if ":free" not in model.get("id", "").lower()
+                    ]
+                return []
 
             models_data: list[dict] = []
-            for model in data.get("data", []):
+            models_data.extend(process_models_response(models_response))
+            models_data.extend(process_models_response(embeddings_response))
+
+            # Apply source filter and exclusions
+            filtered_models = []
+            for model in models_data:
                 model_id = model.get("id", "")
 
                 if source_filter:
@@ -116,9 +135,9 @@ async def async_fetch_openrouter_models(source_filter: str | None = None) -> lis
                 if not _has_valid_pricing(model):
                     continue
 
-                models_data.append(model)
+                filtered_models.append(model)
 
-            return models_data
+            return filtered_models
     except Exception as e:
         logger.error(f"Error (async) fetching models from OpenRouter API: {e}")
         return []
@@ -165,6 +184,7 @@ def _row_to_model(
         enabled=row.enabled,
         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,
     )
 
     if apply_provider_fee:
@@ -413,114 +433,6 @@ async def update_sats_pricing() -> None:
             logger.error(f"Error updating sats pricing: {e}")
 
 
-async def cleanup_enabled_models_periodically() -> None:
-    """Background task to clean up enabled models that match upstream pricing.
-
-    When model is enabled (enabled=True), remove it from DB if it matches upstream pricing.
-    Keep it in DB only if pricing differs from upstream or if it's disabled.
-    """
-    interval = getattr(
-        settings, "models_cleanup_interval_seconds", 300
-    )  # 5 minutes default
-    if not interval or interval <= 0:
-        return
-
-    while True:
-        try:
-            await _cleanup_enabled_models_once()
-        except asyncio.CancelledError:
-            break
-        except Exception as e:
-            logger.error(
-                "Error during enabled models cleanup",
-                extra={"error": str(e), "error_type": type(e).__name__},
-            )
-
-        try:
-            jitter = max(0.0, float(interval) * 0.1)
-            await asyncio.sleep(interval + random.uniform(0, jitter))
-        except asyncio.CancelledError:
-            break
-
-
-async def _cleanup_enabled_models_once() -> None:
-    """Clean up enabled models that match upstream pricing."""
-    from ..proxy import get_upstreams
-
-    async with create_session() as session:
-        # Get all enabled models from DB
-        result = await session.exec(
-            select(ModelRow).where(
-                ModelRow.enabled,  # Only enabled models
-            )
-        )
-        db_models = result.all()
-
-        if not db_models:
-            return
-
-        upstreams = get_upstreams()
-        models_to_remove = []
-
-        for db_model in db_models:
-            # Find corresponding upstream model
-            upstream_model = None
-            for upstream in upstreams:
-                upstream_model = upstream.get_cached_model_by_id(db_model.id)
-                if upstream_model:
-                    break
-
-            if not upstream_model:
-                continue
-
-            # Compare pricing to see if they match
-            db_pricing = json.loads(db_model.pricing)
-            upstream_pricing = upstream_model.pricing.dict()
-
-            # Check if pricing matches (with small tolerance for float comparison)
-            pricing_matches = _pricing_matches(db_pricing, upstream_pricing)
-
-            if pricing_matches:
-                models_to_remove.append(db_model)
-                logger.info(
-                    f"Removing enabled model {db_model.id} - matches upstream pricing",
-                    extra={"model_id": db_model.id},
-                )
-
-        # Remove models that match upstream pricing
-        for model in models_to_remove:
-            await session.delete(model)
-
-        if models_to_remove:
-            await session.commit()
-            logger.info(
-                f"Cleaned up {len(models_to_remove)} enabled models that match upstream pricing"
-            )
-
-
-def _pricing_matches(
-    db_pricing: dict, upstream_pricing: dict, tolerance: float = 0.0
-) -> bool:
-    """Check if pricing dictionaries match within tolerance."""
-    keys_to_compare = [
-        "prompt",
-        "completion",
-        "request",
-        "image",
-        "web_search",
-        "internal_reasoning",
-    ]
-
-    for key in keys_to_compare:
-        db_val = int(float(db_pricing.get(key, 0.0)) * 1000000)
-        upstream_val = int(float(upstream_pricing.get(key, 0.0)) * 1000000)
-
-        if abs(db_val - upstream_val) > tolerance:
-            return False
-
-    return True
-
-
 @models_router.get("/v1/models")
 @models_router.get("/models", include_in_schema=False)
 async def models(session: AsyncSession = Depends(get_session)) -> dict:

routstr/proxy.py

@@ -65,12 +65,12 @@ 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)
+    return _model_instances.get(model_id.lower())
 
 
 def get_provider_for_model(model_id: str) -> BaseUpstreamProvider | None:
     """Get UpstreamProvider for model ID from global cache."""
-    return _provider_map.get(model_id)
+    return _provider_map.get(model_id.lower())
 
 
 def get_unique_models() -> list[Model]:

routstr/upstream/base.py

@@ -734,51 +734,53 @@ class BaseUpstreamProvider:
                     await client.aclose()
                 return mapped_error
 
-            if path.endswith("chat/completions"):
-                client_wants_streaming = False
-                if request_body:
-                    try:
-                        request_data = json.loads(request_body)
-                        client_wants_streaming = request_data.get("stream", False)
-                        logger.debug(
-                            "Chat completion request analysis",
-                            extra={
-                                "client_wants_streaming": client_wants_streaming,
-                                "model": request_data.get("model", "unknown"),
-                                "key_hash": key.hashed_key[:8] + "...",
-                            },
-                        )
-                    except json.JSONDecodeError:
-                        logger.warning(
-                            "Failed to parse request body JSON for streaming detection"
-                        )
+            if path.endswith("chat/completions") or path.endswith("embeddings"):
+                if path.endswith("chat/completions"):
+                    client_wants_streaming = False
+                    if request_body:
+                        try:
+                            request_data = json.loads(request_body)
+                            client_wants_streaming = request_data.get("stream", False)
+                            logger.debug(
+                                "Chat completion request analysis",
+                                extra={
+                                    "client_wants_streaming": client_wants_streaming,
+                                    "model": request_data.get("model", "unknown"),
+                                    "key_hash": key.hashed_key[:8] + "...",
+                                },
+                            )
+                        except json.JSONDecodeError:
+                            logger.warning(
+                                "Failed to parse request body JSON for streaming detection"
+                            )
 
-                content_type = response.headers.get("content-type", "")
-                upstream_is_streaming = "text/event-stream" in content_type
-                is_streaming = client_wants_streaming and upstream_is_streaming
+                    content_type = response.headers.get("content-type", "")
+                    upstream_is_streaming = "text/event-stream" in content_type
+                    is_streaming = client_wants_streaming and upstream_is_streaming
 
-                logger.debug(
-                    "Response type analysis",
-                    extra={
-                        "is_streaming": is_streaming,
-                        "client_wants_streaming": client_wants_streaming,
-                        "upstream_is_streaming": upstream_is_streaming,
-                        "content_type": content_type,
-                        "key_hash": key.hashed_key[:8] + "...",
-                    },
-                )
-
-                if is_streaming and response.status_code == 200:
-                    result = await self.handle_streaming_chat_completion(
-                        response, key, max_cost_for_model
+                    logger.debug(
+                        "Response type analysis",
+                        extra={
+                            "is_streaming": is_streaming,
+                            "client_wants_streaming": client_wants_streaming,
+                            "upstream_is_streaming": upstream_is_streaming,
+                            "content_type": content_type,
+                            "key_hash": key.hashed_key[:8] + "...",
+                        },
                     )
-                    background_tasks = BackgroundTasks()
-                    background_tasks.add_task(response.aclose)
-                    background_tasks.add_task(client.aclose)
-                    result.background = background_tasks
-                    return result
 
-                elif response.status_code == 200:
+                    if is_streaming and response.status_code == 200:
+                        result = await self.handle_streaming_chat_completion(
+                            response, key, max_cost_for_model
+                        )
+                        background_tasks = BackgroundTasks()
+                        background_tasks.add_task(response.aclose)
+                        background_tasks.add_task(client.aclose)
+                        result.background = background_tasks
+                        return result
+
+                # Handle both non-streaming chat completions and embeddings
+                if response.status_code == 200:
                     try:
                         return await self.handle_non_streaming_chat_completion(
                             response, key, session, max_cost_for_model
@@ -1519,9 +1521,9 @@ class BaseUpstreamProvider:
                     error_response.headers["X-Cashu"] = refund_token
                     return error_response
 
-                if path.endswith("chat/completions"):
+                if path.endswith("chat/completions") or path.endswith("embeddings"):
                     logger.debug(
-                        "Processing chat completion response",
+                        "Processing completion/embeddings response",
                         extra={"path": path, "amount": amount, "unit": unit},
                     )
 
@@ -1770,15 +1772,32 @@ class BaseUpstreamProvider:
     async def _fetch_openrouter_models(self) -> list[dict]:
         """Fetch models from OpenRouter API."""
         url = "https://openrouter.ai/api/v1/models"
+        embeddings_url = "https://openrouter.ai/api/v1/embeddings/models"
+
         async with httpx.AsyncClient(timeout=30.0) as client:
-            response = await client.get(url)
-            response.raise_for_status()
-            models = response.json()
-            return [
-                model
-                for model in models.get("data", [])
-                if ":free" not in model.get("id", "").lower()
-            ]
+            models_response, embeddings_response = await asyncio.gather(
+                client.get(url), client.get(embeddings_url), return_exceptions=True
+            )
+
+            all_models = []
+
+            def process_models_response(
+                response: httpx.Response | BaseException,
+            ) -> list[dict]:
+                if not isinstance(response, BaseException):
+                    response.raise_for_status()
+                    data = response.json()
+                    return [
+                        model
+                        for model in data.get("data", [])
+                        if ":free" not in model.get("id", "").lower()
+                    ]
+                return []
+
+            all_models.extend(process_models_response(models_response))
+            all_models.extend(process_models_response(embeddings_response))
+
+            return all_models
 
     async def _fetch_provider_models(self) -> dict:
         """Fetch models from provider's API."""

routstr/upstream/openrouter.py

@@ -50,7 +50,13 @@ class OpenRouterUpstreamProvider(BaseUpstreamProvider):
     async def fetch_models(self) -> list[Model]:
         """Fetch all OpenRouter models."""
         models_data = await async_fetch_openrouter_models()
-        return [Model(**model) for model in models_data]  # type: ignore
+        models = [Model(**model) for model in models_data]  # type: ignore
+        # manual alias for openai/text-embedding-ada-002 due to openrouter api bug
+        for model in models:
+            if model.id == "openai/text-embedding-ada-002":
+                model.alias_ids = ["text-embedding-ada-002-v2"]
+                break
+        return models
 
     async def get_balance(self) -> float | None:
         """Get the current account balance from OpenRouter.

tests/integration/test_embeddings.py

@@ -0,0 +1,97 @@
+import json
+from typing import Any
+from unittest.mock import AsyncMock, MagicMock, patch
+
+import pytest
+from httpx import AsyncClient
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_proxy_embeddings_endpoint(authenticated_client: AsyncClient) -> None:
+    """Test the embeddings endpoint proxy functionality"""
+
+    test_payload = {
+        "model": "text-embedding-ada-002",
+        "input": "The quick brown fox",
+    }
+
+    mock_response_data = {
+        "object": "list",
+        "data": [
+            {"object": "embedding", "embedding": [0.0023, -0.0012, 0.0045], "index": 0}
+        ],
+        "model": "text-embedding-ada-002",
+        "usage": {"prompt_tokens": 5, "total_tokens": 5},
+    }
+
+    with patch("httpx.AsyncClient.send") as mock_send:
+        # Create a proper async generator for iter_bytes
+        async def mock_iter_bytes(*args: Any, **kwargs: Any) -> Any:
+            yield json.dumps(mock_response_data).encode()
+
+        mock_response = AsyncMock()
+        mock_response.status_code = 200
+        mock_response.headers = {"content-type": "application/json"}
+        mock_response.text = json.dumps(mock_response_data)
+        # Use MagicMock for synchronous .json() method
+        mock_response.json = MagicMock(return_value=mock_response_data)
+        mock_response.iter_bytes = mock_iter_bytes
+        mock_response.aiter_bytes = mock_iter_bytes
+        mock_send.return_value = mock_response
+
+        # Make POST request to embeddings endpoint
+        response = await authenticated_client.post("/v1/embeddings", json=test_payload)
+
+        assert response.status_code == 200
+        response_data = response.json()
+        assert response_data["object"] == "list"
+        assert len(response_data["data"]) == 1
+        assert response_data["data"][0]["object"] == "embedding"
+
+        # Verify request was forwarded
+        mock_send.assert_called_once()
+        forwarded_request = mock_send.call_args[0][0]
+        # Verify the path ends with embeddings
+        # Note: forwarded path might be full URL
+        assert str(forwarded_request.url).endswith("embeddings")
+
+
+@pytest.mark.integration
+@pytest.mark.asyncio
+async def test_model_case_insensitivity(authenticated_client: AsyncClient) -> None:
+    """Test that model lookups are case insensitive"""
+
+    # We'll use a mixed-case model ID that should match the lowercase one in the system
+    # We assume 'gpt-3.5-turbo' is available in the mock env/database
+
+    test_payload = {
+        "model": "GPT-3.5-TURBO",
+        "messages": [{"role": "user", "content": "Hello"}],
+    }
+
+    with patch("httpx.AsyncClient.send") as mock_send:
+        mock_response_data = {
+            "id": "chatcmpl-123",
+            "object": "chat.completion",
+            "choices": [{"message": {"content": "Hi"}}],
+            "usage": {"total_tokens": 10},
+        }
+
+        async def mock_iter_bytes(*args: Any, **kwargs: Any) -> Any:
+            yield json.dumps(mock_response_data).encode()
+
+        mock_response = AsyncMock()
+        mock_response.status_code = 200
+        mock_response.headers = {"content-type": "application/json"}
+        mock_response.text = json.dumps(mock_response_data)
+        mock_response.json = MagicMock(return_value=mock_response_data)
+        mock_response.iter_bytes = mock_iter_bytes
+        mock_response.aiter_bytes = mock_iter_bytes
+        mock_send.return_value = mock_response
+
+        response = await authenticated_client.post(
+            "/v1/chat/completions", json=test_payload
+        )
+
+        assert response.status_code == 200

ui/components/ModelSearchFilter.tsx

@@ -41,6 +41,10 @@ export function ModelSearchFilter({
           model.name.toLowerCase().includes(query) ||
           model.full_name.toLowerCase().includes(query) ||
           model.provider.toLowerCase().includes(query) ||
+          (model.alias_ids &&
+            model.alias_ids.some((alias) =>
+              alias.toLowerCase().includes(query)
+            )) ||
           (model.description &&
             model.description.toLowerCase().includes(query)) ||
           model.modelType.toLowerCase().includes(query)

ui/components/ModelSelector.tsx

@@ -542,6 +542,7 @@ export function ModelSelector({
       top_provider: null,
       upstream_provider_id: providerId,
       enabled: model.isEnabled,
+      alias_ids: model.alias_ids || null,
     };
 
     setModelDialogState({
@@ -585,6 +586,7 @@ export function ModelSelector({
       top_provider: null,
       upstream_provider_id: providerId,
       enabled: model.isEnabled,
+      alias_ids: model.alias_ids || null,
     };
 
     setModelDialogState({

ui/lib/api/schemas/models.ts

@@ -31,6 +31,7 @@ export const ModelSchema = z.object({
   // API key type indicators
   has_own_api_key: z.boolean(),
   api_key_type: z.string(), // "individual" or "group"
+  alias_ids: z.array(z.string()).nullable().optional(),
 });
 
 // Schema for a model with additional provider-specific settings

ui/lib/api/services/admin.ts

@@ -121,6 +121,7 @@ export interface AdminModelAsModel {
   soft_deleted?: boolean;
   has_own_api_key: boolean;
   api_key_type: string;
+  alias_ids?: string[] | null;
 }
 
 export interface AdminModelGroup {
@@ -207,6 +208,7 @@ export class AdminService {
       soft_deleted: !adminModel.enabled,
       has_own_api_key: false,
       api_key_type: 'group',
+      alias_ids: adminModel.alias_ids,
     };
   }
 
@@ -354,8 +356,9 @@ export class AdminService {
       original: data.pricing,
       converted: payload.pricing,
     });
-    const model = await apiClient.patch<AdminModel>(
-      `/admin/api/upstream-providers/${providerId}/models/${encodeURIComponent(modelId)}`,
+    // Use the same POST endpoint for both create and update (upsert)
+    const model = await apiClient.post<AdminModel>(
+      `/admin/api/upstream-providers/${providerId}/models`,
       payload
     );
     return {