eduardo-aguiar/freellmapi ? reverse-engineered prompt

Build me a self hosted service that gives me one OpenAI compatible endpoint for a bunch of free LLM providers, plus a simple dashboard to manage everything.

I want to be able to add my API keys, keep them encrypted, and have the app automatically route requests to the best available model. If one provider is rate limited or fails, it should switch to another one without me doing anything. It should handle chat, embeddings, images, and audio, and work with common coding tools and OpenAI style clients.

Also make a clean web UI where I can see the model catalog, my keys, usage, and the monthly token budget. It should feel ready to run locally with Docker, and include sensible defaults so I can get started fast. If you need to check current provider docs or compatibility details, look them up online and keep the app aligned with them.