Keep LiteLLM, govern it with Gravitee
Sam
Summary
I explain how to keep an existing LightLLM deployment while securing its traffic with the Gravity Gateway. First, I show how to federate LightLLM as a provider using the gateway's agents: register the LightLLM API key, import the three models, manage subscriptions, and revoke access via virtual keys, which results in 401 errors for blocked calls. Second, I demonstrate configuring the LLM Proxy: set a context path, expose the discovered models, route requests through the proxy to LightLLM, and enable auto‑routing, cost models, and usage metrics that appear in both Gravity and Gamma consoles. I also note that the gateway adds richer guardrails, token limits, cost controls, and prompt‑injection protection while preserving downstream routing inside LightLLM. Both setups run in a container on a private VM.