Using Internally Hosted LLM and Embedding Model

[sssaha1989]

Hello
I am running into issues to use the LLM and Embedding models hosted internally by our organization. I have setup the environment varialbes

LLM_API_KEY="dummy-key"
LLM_MODEL="google/gemma-3-27b-it"
LLM_ENDPOINT="http://xx.com:8000/v1"
LLM_PROVIDER="custom"

EMBEDDING_MODEL="thenlper/gte-large"
EMBEDDING_ENDPOINT="http://xx.com:8003/v1"
EMBEDDING_DIMENSIONS=512
EMBEDDING_MAX_TOKENS=512
EMBEDDING_PROVIDER="custom"
EMBEDDING_API_KEY="dummy-key"

However - it will always complain something like this

litellm.BadRequestError: LLM Provider NOT provided.

Now I tried litellm directly using their VLLM configurations - and that code worked. If I provide the same parameters that I use in my LITELLM code while running the cognee code - it will throw an error.

I tried without the v1 - but that gives me the same error.

Is this even possible using Cognee?