MiniMLX: Run AI models on your Mac. Fast and simple.

Hi Product Hunt! 👋

I built MiniMLX to make running local AI models on a Mac feel simpler and more approachable.

I wanted one place to discover models, download them, start an inference service, and see how everything is performing—without switching between terminal commands, model repositories, and monitoring tools.

MiniMLX is a native macOS app built for Apple Silicon. You can:

• Discover, download, and manage a wide range of popular MLX models from Hugging Face and ModelScope

• Run compatible models locally on your Mac

• Connect your apps through an OpenAI-compatible API

• Measure generation speed and monitor memory and GPU usage

• Control the service from the macOS menu bar

MiniMLX also supports MTP acceleration for compatible models and provides options for running certain larger models on Macs with less available memory.

Projects like oMLX inspired me to explore what's possible with local inference on Apple Silicon. With MiniMLX, my focus is a simple desktop experience that makes finding, running, and managing models feel natural on a Mac.

MiniMLX is open source and free to use. This is our first release, and I'd love to hear about your experience—especially which models you run and what would make the workflow better for you.

GitHub: https://github.com/shijizhi/MiniMLX