[Bug]: OpenRouter support for embeddings

[conext-noc]

What happened?

Support for OpenRouter embeddings in LiteLLM is crucial because, currently, there are bugs or compatibility limitations that prevent their explicit configuration and direct use.

Relevant log output

if use openrouter provider and custom model name:
1) google/gemini-embedding-001 - Error is , LLM Provider NOT provided. Pass in the LLM provider you are trying to call. You passed model=google/gemini-embedding-001 Pass model as E.g. For 'Huggingface' inference endpoints pass in `completion(model='huggingface/starcoder',..)` Learn more: https://docs.litellm.ai/docs/providers

2) openrouter/google/gemini-embedding-001 - Error is, Unmapped LLM provider for this endpoint. You passed model=google/gemini-embedding-001, custom_llm_provider=openrouter. Check supported provider and route: https://docs.litellm.ai/docs/providers

Are you a ML Ops Team?

No

What LiteLLM version are you on ?

v1.79.3-stable.gemini3

Twitter / LinkedIn details

No response

[chaserhkj]

Can confirm this, was just about to submit my own issue report when I see this.

LiteLLM proxy is not routing embedding requests to openrouter endpoint.

I am using docker image main-latest nightly build, I have in my config:

model_list:
  - model_name: test-model
    litellm_params:
      model: openrouter/qwen/qwen3-embedding-8b

But:

$ curl http://192.168.233.1:4000/v1/embeddings \
  -H "Content-Type: application/json" \
  -d '{"input": "test sentence", "model": "test-model"}'
{
  "error": {
    "message": "litellm.BadRequestError: Unmapped LLM provider for this endpoint. You passed model=qwen/qwen3-embedding-8b, custom_llm_provider=openrouter. Check supported provider and route: https://docs.litellm.ai/docs/providers. Received Model Group=test-model\nAvailable Model Group Fallbacks=None",
    "type": null,
    "param": null,
    "code": "400"
  }
}

And I directly tested the openrouter endpoint with my API key just in case and it was fine:

$ curl https://openrouter.ai/api/v1/embeddings \
-H "Authorization: Bearer <Same token as in litellm.yaml>" \
-H "Content-Type: application/json" \
-d '{"input": "test sentence", "model": "qwen/qwen3-embedding-8b"}'

{
  "object": "list",
  "data": [
    {
      "object": "embedding",
      "embedding": [<omitted embedding values>]
      "index": 0
    }
  ],
  "model": "Qwen/Qwen3-Embedding-8B",
  "usage": {
    "prompt_tokens": 3,
    "total_tokens": 3,
    "cost": 1.2E-7
  },
  "provider": "SiliconFlow",
  "id": "<redacted>"
}

Seems to be an issue with LiteLLM end.

[Magolor]

Can confirm this, was just about to submit my own issue report when I see this.

LiteLLM proxy is not routing embedding requests to openrouter endpoint.

I am using docker image main-latest nightly build, I have in my config:

model_list:

• model_name: test-model
  litellm_params:
  model: openrouter/qwen/qwen3-embedding-8b
  But:

$ curl http://192.168.233.1:4000/v1/embeddings
-H "Content-Type: application/json"
-d '{"input": "test sentence", "model": "test-model"}'
{
"error": {
"message": "litellm.BadRequestError: Unmapped LLM provider for this endpoint. You passed model=qwen/qwen3-embedding-8b, custom_llm_provider=openrouter. Check supported provider and route: https://docs.litellm.ai/docs/providers. Received Model Group=test-model\nAvailable Model Group Fallbacks=None",
"type": null,
"param": null,
"code": "400"
}
}
And I directly tested the openrouter endpoint with my API key just in case and it was fine:

$ curl https://openrouter.ai/api/v1/embeddings
-H "Authorization: Bearer "
-H "Content-Type: application/json"
-d '{"input": "test sentence", "model": "qwen/qwen3-embedding-8b"}'

{
"object": "list",
"data": [
{
"object": "embedding",
"embedding": []
"index": 0
}
],
"model": "Qwen/Qwen3-Embedding-8B",
"usage": {
"prompt_tokens": 3,
"total_tokens": 3,
"cost": 1.2E-7
},
"provider": "SiliconFlow",
"id": ""
}
Seems to be an issue with LiteLLM end.

Also confirmed on litellm 1.80.7 and 1.80.9 with openai embedding model:

19:52:07 - LiteLLM:DEBUG: utils.py:381 - 

19:52:07 - LiteLLM:DEBUG: utils.py:381 - Request to litellm:
19:52:07 - LiteLLM:DEBUG: utils.py:381 - litellm.embedding(input=['hello'], seed=42, timeout=120, api_key=..., api_base='https://openrouter.ai/api/v1', model='openrouter/openai/text-embedding-3-small')
19:52:07 - LiteLLM:DEBUG: utils.py:381 - 

19:52:07 - LiteLLM:DEBUG: litellm_logging.py:509 - self.optional_params: {}
19:52:07 - LiteLLM:DEBUG: utils.py:381 - SYNC kwargs[caching]: False; litellm.cache: None; kwargs.get('cache')['no-cache']: False
19:52:07 - LiteLLM:DEBUG: litellm_logging.py:509 - self.optional_params: {'seed': 42}
19:52:07 - LiteLLM:DEBUG: litellm_logging.py:1121 - RAW RESPONSE:
litellm.BadRequestError: Unmapped LLM provider for this endpoint. You passed model=openai/text-embedding-3-small, custom_llm_provider=openrouter. Check supported provider and route: https://docs.litellm.ai/docs/providers


19:52:07 - LiteLLM:DEBUG: litellm_logging.py:2692 - Logging Details LiteLLM-Failure Call: []
Error during embedding: litellm.BadRequestError: Unmapped LLM provider for this endpoint. You passed model=openai/text-embedding-3-small, custom_llm_provider=openrouter. Check supported provider and route: https://docs.litellm.ai/docs/providers

Meanwhile, ollama as an embedding model backend works, so it is a litellm-openrouter compatibility exclusive issue.

[CesarSanch1998]

I also have the same problem when trying to use openrouter embeddings, the only way it allowed me is by selecting all the models using openrouter/*, but this selects models at random every time and it doesn't help me.