ddalcu/mlx-serve

Native LLM inference server for Apple Silicon. OpenAI + Anthropic API compatible. No Python. Zig backend, Swift frontend macOS app with chat, music, voice, video generation.

Latest release v26.10.3 — v26.10.3 — New Models, Transcripts, RDMA, Faster · 22 Oct 2026

5 files for Claude Code, Codex and OpenCode: bench, release, mlx-serve CLAUDE.md, pr-charts and 1 more — 7,742 tokens loaded in every session.

Open on GitHub mlxserve.com

Instructions

mlx-serve CLAUDE.md CLAUDE.md A 5 tok
mlx-serve AGENTS.md AGENTS.md C 7,737 tok

Skills

bench .claude/skills/bench/SKILL.md A 53 tok
release .claude/skills/release/SKILL.md A 42 tok
pr-charts .claude/skills/pr-charts/SKILL.md A 66 tok

These files are ddalcu/mlx-serve's own configuration — they tell Claude Code, Codex and OpenCode how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.