songquanpeng/one-api ? reverse-engineered prompt

Build me a self hosted AI API management system that lets me use one standard OpenAI style interface to connect to a bunch of model providers like OpenAI, Azure, Claude, Gemini, DeepSeek, Qwen, and others.

I want a clean web dashboard where I can add and organize API keys, create user tokens with expiry and usage limits, set allowed models and IP ranges, manage channels, see balance details, and route requests through load balancing with automatic retries. It should support streaming responses, model mapping, shared quotas, and a simple way to generate and redeem recharge codes. Please include user login and admin access, with a default root account on first setup, and make it easy to change branding like the site name, logo, and footer.

Make it easy to deploy with Docker or a single binary, save data locally, and be ready to run out of the box. If you need current provider docs or setup details, look them up online and then wire things up sensibly.