
by BerriAI
LiteLLM is an open-source, self-hosted AI gateway that puts many model providers behind one OpenAI-compatible API. It adds virtual keys, spend tracking, routing and an admin UI.
APIFreemium
Step 1: Start the gateway with the quickstart script or the Docker Compose file; it runs on port 4000 with a Postgres database.
Step 2: Log in to the Admin UI with the master key you generated.
Step 3: Add a model by choosing a provider and pasting its API key, then test the connection.
Step 4: Create a virtual key with the budget and limits you want.
Step 5: Point any OpenAI SDK at the gateway URL using the virtual key and send requests.
Step 6: Watch spend and request logs in the Admin UI.
Yes. The site describes it as open source and self-hosted, and you can run it with a single command or Docker.
The docs state that no data or telemetry is stored on LiteLLM servers when you self-host. Usage telemetry is off by default.
Any client that works with the OpenAI API works with the gateway by changing the base URL and key.
The site lists SSO login scoped by team, project or app, and the docs mention audit logs.
TwelveLabsVideo understanding API for searching, analyzing, embedding and generating insights from video content.