# MLX Omni Server

A local inference server for Apple Silicon that serves MLX models behind both OpenAI-compatible and Anthropic-compatible API endpoints, so existing SDK-based clients can point at it as a drop-in local replacement.

MLX Omni Server is a local AI inference server built on Apple's MLX framework for M-series chips, exposing both an OpenAI-compatible API and an Anthropic-compatible API over the same running server. All processing happens on the machine running it, so an existing OpenAI or Anthropic SDK client can be pointed at the local server with no other code changes.

Supported endpoints cover chat completions with tools, streaming and structured output, text-to-speech, speech-to-text, image generation, text embeddings and model management, reachable at OpenAI-style paths under /v1 and Anthropic-style paths under /anthropic/v1, including a messages endpoint with tools, streaming and a thinking mode.

It installs as a Python package and starts as a single command, with configurable options such as the listening port.

Website: https://deepwiki.com/madroidmaq/mlx-omni-server/1-overview