Analysis updated 2026-05-18
Point an existing OpenAI-compatible app at a free model with no code changes.
Self-host a lightweight AI proxy on a Debian, Ubuntu, RHEL, or Alpine server.
Run the proxy as a background service that restarts automatically on reboot.
Restrict the proxy to localhost only for private, single-machine use.
| xuomen/mimo-free-proxy | ai-engineer-skool/zen-agentic-engineer-config | ali-expandings/mactune | |
|---|---|---|---|
| Stars | 16 | 16 | 16 |
| Language | Shell | Shell | Shell |
| Setup difficulty | easy | easy | easy |
| Complexity | 2/5 | 2/5 | 1/5 |
| Audience | developer | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Needs a Linux server and root access, the one-line curl installer handles dependencies and the service manager automatically.
mimo-free-proxy is a single-script tool that turns Xiaomi's MiMoCode free AI channel into a standard OpenAI-compatible API endpoint on a Linux server. Any app or library that already knows how to talk to OpenAI's chat API can be pointed at this proxy instead, giving access to Xiaomi's free model without code changes on the client side. Installation is a one-line command that downloads and runs a setup script. The script detects which Linux distribution you're on, installs any missing dependencies, registers the proxy as a background service using systemd or OpenRC (the two common service managers on Linux), and prints three pieces of information: the API address, an auto-generated API key, and the model name. From that point the service runs automatically and restarts itself when the machine reboots. The underlying model is called mimo-auto, described as a reasoning model from Xiaomi. The proxy handles authentication automatically, refreshing its internal session tokens before they expire and retrying if they are rejected. There is no database, no Node.js, and no other framework required beyond Python 3. A few limitations are noted in the README. Rate limiting is enforced per server IP address rather than per API key, so running multiple keys on the same server does not increase available capacity. The service opens a network port (8788 by default), so if you run it on a public server the README advises keeping the API key private. There is an option to restrict the service to localhost only for local use. For larger public deployments the README suggests adding a reverse proxy with rate limiting in front of it.
A one-line install script that turns Xiaomi's free MiMoCode AI channel into a standard OpenAI-compatible API endpoint on your own Linux server.
Mainly Shell. The stack also includes Python, Shell, systemd.
MIT license: use, modify, and distribute freely, including commercially, as long as you keep the copyright notice.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.