Pico MLX Server

Pico MLX Server: The command-line-free, local LLM server for Apple Silicon, providing an OpenAI-compatible API for MLX-based models.

Pico MLX Server delivers a high-performance, local AI solution built on Apple’s MLX framework, optimized exclusively for M-series chips (M1 or later). It provides an OpenAI-compatible API, specifically the POST /v1/completions endpoint, allowing integration with existing chat clients. This server simplifies deployment with a GUI and automated setup, eliminating the need for complex command-line configuration. Users can run over 300 models locally—including Meta Llama and Google Gemma—leveraging the speed of Apple Silicon for performance up to 5 times faster than competing local AI applications.

https://github.com/PicoMLX/PicoMLXServer