Skip to content

LiteLLM

A LiteLLM proxy exposing an OpenAI-compatible endpoint, selected with LLM_PROVIDER=litellm, used to extract facts from conversations.

Enable it

bash
LLM_PROVIDER=litellm
OPENAI_BASE_URL=http://localhost:4000
OPENAI_API_KEY=<litellm-proxy-key>
OPENAI_LLM_MODEL=<model-name-configured-in-litellm>

Configuration

VariableDefaultDescription
LLM_PROVIDERopenaiSet to litellm.
OPENAI_BASE_URLhttps://api.openai.com/v1Point at your LiteLLM proxy: http://localhost:4000.
OPENAI_API_KEY(none)Proxy key, sent as a bearer token.
OPENAI_LLM_MODELgpt-4.1-nanoA model name routed by your LiteLLM config.

Notes

  • OpenAI-compatible: the provider reuses the OpenAI request handling and only differs by base URL, key, and model.
  • The provider also reads LITELLM_BASE_URL (default http://localhost:4000) and LITELLM_API_KEY as its own fallbacks.
  • LiteLLM itself fans out to whatever downstream models its proxy config defines.

The memory layer for AI agents.