MTPLX is an open-source LLM inference tool written in Python, released under the Apache-2.0 license. The repository has collected 2.6k stars and 202 forks; the latest push was in October 2026. Topics include anthropic-compatible, apple-silicon and claude-code.
“The fastest way to run Qwen 3.8 Flash Next, Qwen 3.8 27B and Ternary Bonsai 2 27B on a Mac: 125 tok/s in OpenCode on an M5 Max, and a 27B model on 16 GB Macs. Native MTP speculative decoding on Apple Silicon, exact at any temperature. OpenAI and Anthropic compatible local server.”
— youssofal, project description on GitHub
Details
| Category | Local LLMs & Inference |
|---|---|
| GitHub stars | 2,600 |
| Forks | 202 |
| Language | Python |
| License | Apache-2.0 |
| Last updated | |
| Owner | youssofal |
| Website | mtplx.com |
| Repository | youssofal/MTPLX |
Tags
anthropic-compatibleapple-siliconclaude-codeflash-nextinference-enginellm-inferencelocal-ailocal-llm
Repository facts from the GitHub public API, refreshed October 18, 2026. The quoted tagline is written by the project owner; the summary above is generated from public facts.