Rapid-MLX
Rapid-MLX is an open-source (Apache 2.0) OpenAI- and Anthropic-compatible LLM inference server and Mac app for Apple Silicon, built on MLX, focused on reliable
What is Rapid-MLX?
Rapid-MLX is an open-source (Apache 2.0) OpenAI- and Anthropic-compatible LLM inference server and Mac app for Apple Silicon, built on MLX, focused on reliable tool calling for coding agents. Release-gated with Claude Code, Codex CLI, Aider, Hermes and DeepSeek Harness.
Overview of Rapid-MLX
What it is
Rapid-MLX is an open-source (Apache 2.0) OpenAI- and Anthropic-compatible LLM inference server and Mac app for Apple Silicon, built on MLX. It is focused on reliable tool calling for coding agents. The project is release-gated with Claude Code, Codex CLI, Aider, Hermes and DeepSeek Harness.
Key features
- OpenAI-compatible (
/v1/chat/completions,/v1/responses) and Anthropic Messages (/v1/messages) API endpoints - Up to 4× faster decode than Apple's MLX (mlx-lm) and faster than oMLX and LM Studio on tested models (e.g., Qwen3.5-9B, Qwen3.8-27B, Qwen3.6-35B) - 27 tool-call parser modules including an auto-detect fallback - Continuous batching for concurrent requests - Radix prefix cache in memory with disk persistence and quantized KV cache - Desktop app (macOS) for chat, model management, vision, files, voice, and image generation - CLI and server with guided installer and Homebrew support
Use cases
- Local LLM inference for coding agents (Claude Code, Codex CLI, Aider, etc.)
- Chat and model serving for other applications via OpenAI/Anthropic-compatible endpoints - Community benchmarking to share standardized performance results
Getting started
bash # Install via Homebrew brew install rapid-mlx # Or use the guided installer curl -fsSL https://rapidmlx.com/install.sh | bash # Quick start: chat with a model rapid-mlx chat # Serve the model for other apps rapid-mlx serve qwen3.5-4b-4bit # Wire up a coding agent (e.g., Claude Code) rapid-mlx launch claude-code
License and pricing
Licensed under Apache 2.0. The project is open source and free to use.
Based on the project's public README and repository metadata.