One endpoint in front of every LLM backend you run

Admins register hosted models, Nebius Token Factory endpoints or OpenRouter instances and map model names to them. Users register, mint their own API keys, and never see your upstream credentials.

Get an API key

Backend registry

Add any OpenAI-compatible endpoint with its own base URL, auth header and credential.

Per-user keys

Each user mints and revokes their own router keys — independent of your backend auth.

LangSmith LLMOps

Every routed call is traced with latency, tokens and errors into your LangSmith project.