Analysis updated 2026-05-18
Point a coding assistant like Claude Code or GitHub Copilot at a local server that forwards requests to MiMo's free web tier.
Use MiMo's free chat quota through the standard OpenAI or Anthropic API format instead of MiMo's rate-limited official API.
Configure multiple MiMo accounts so the proxy rotates between them to spread request load.
Monitor token usage, request counts, and model distribution from a built-in web dashboard.
| wtz44/mimo-free-api | duckbugio/flock | glockinhand/guns.lol-view-bot | |
|---|---|---|---|
| Stars | 38 | 36 | 40 |
| Language | Go | Go | Go |
| Setup difficulty | moderate | hard | moderate |
| Complexity | 3/5 | 4/5 | 3/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires manually copying session cookies from a logged-in browser tab, and these cookies expire and must be refreshed periodically.
MiMo Free API is a local server that sits between AI tools you already use and the Xiaomi MiMo AI web interface. Xiaomi offers a MiMo AI chat product at aistudio.xiaomimimo.com, and the web version is free to use, but there is no official API for connecting it to other software. This project fills that gap by running a small server on your machine that speaks the same API format as OpenAI and Anthropic, so any tool that knows how to talk to those services can now talk to MiMo instead. The practical effect is that coding assistants, chat frontends, and agent frameworks (the README lists Claude Code, Codex, GitHub Copilot, and several others) can be pointed at http://localhost:8080 and will work as if they were talking to a standard AI API, while the requests are actually going to MiMo's free web tier. To connect MiMo to the proxy, you log into the MiMo website in your browser, open the browser's developer tools network panel, send a chat message, and copy three cookie values from the outgoing request: a service token, a user ID, and a third session value. These go into a config file. The cookies expire after some days or weeks, so you need to repeat this step when they stop working. The server supports both OpenAI-style requests (at /v1/chat/completions) and Anthropic-style requests (at /v1/messages). It handles tool calling in both formats, multi-modal inputs such as images and audio files, and can maintain conversation context across multiple turns. You can configure multiple MiMo accounts and the proxy will rotate between them to spread the load. A web dashboard at the same local address shows statistics: how many tokens have been used, how many requests have been made, and which models are being called. The two available models are mimo-v2.5 (general, handles images, audio, and files) and mimo-v2.5-pro (text and images only, but with stronger reasoning). The server is written in Go and compiles to a single binary with the React dashboard embedded inside it. MIT licensed. The README is written in Chinese with an English section at the end.
A local proxy server that turns Xiaomi's free MiMo AI web chat into an OpenAI and Anthropic compatible API for coding tools and agents.
Mainly Go. The stack also includes Go, React, TypeScript.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.