Holy crap, this tool is made for freeloading LLMs, perfect for college students Someone combined the free tiers of over a dozen AI platforms into a pool of free models that automatically tests speed, selects models, and switches to the next one if one goes down. Kimi, DeepSeek, Qwen, GLM, GPT-OSS, Gemma—they’re all included. The project is called *free-coding-models*. Before, if you wanted to freeload models for coding, you had to register for Groq, Cerebras, NVIDIA, OpenRouter, Google AI, Mistral… then manually find models, test speeds, copy API addresses, and mess around with configuring tools like OpenCode, Goose, Aider, or Cline. This tool takes care of all that messy work for you. Once you start it, it’ll detect which models you can use in real time, test latency, stability, and disconnection rates, and then rank the best ones available. Pick one, hit Enter, and it’ll automatically write the config into your coding tool and launch it. It already supports 20+ dev tools, including OpenCode, Goose, Aider, Crush, Continue, Cline, OpenHands, Qwen Code, Copilot, and Xcode. The best part? It even has a local Router. You don’t even need to pick a model yourself—your Coding Agent will always connect to *fcm*. If the free models hit 429, 500, 529 errors, overload, or go down, it’ll just switch to the next one. It’s like turning all the free tiers into one big free model cluster. The project was only created this February, and it already has 2,200+ stars. https://(x.com)/om_patel5/status/2040298602069639208/video/1