Skip to content

OpenAI

OpenAI chat-completions, selected with LLM_PROVIDER=openai, used to extract facts from conversations. This is the default provider.

Enable it

bash
LLM_PROVIDER=openai
OPENAI_API_KEY=sk-...
OPENAI_BASE_URL=https://api.openai.com/v1
OPENAI_LLM_MODEL=gpt-4.1-nano

Configuration

VariableDefaultDescription
LLM_PROVIDERopenaiSet to openai.
OPENAI_API_KEY(none)API key sent as a bearer token. Calls fail at runtime if unset.
OPENAI_BASE_URLhttps://api.openai.com/v1Chat-completions base URL.
OPENAI_LLM_MODELgpt-4.1-nanoModel used for extraction.

Notes

  • gpt-4.1-nano is the default: cheap and fast for high-volume fact extraction.
  • Requests run at temperature=0.1 with a 2000-token response cap.
  • Any OpenAI-compatible endpoint works here by changing OPENAI_BASE_URL; the other OpenAI-compatible providers are thin presets over this same mechanism.

The memory layer for AI agents.