octopus

Why we included this project

People who work with several LLM providers usually end up managing a different key, endpoint, and dashboard for each one. Octopus is a self-hosted gateway that pulls those together: you can attach multiple API keys to a single channel, and it routes each request to the endpoint with the shortest delay while spreading load across the pool. It also converts between OpenAI Chat, OpenAI Responses, and Anthropic API formats, so a tool written against one provider can talk to the others without code changes. The web panel tracks requests, token use, and costs, and can sync model lists and pricing automatically. Deployment is a single binary or a Docker container, so you don't need heavy infrastructure to run it.