mtplx
mtplx is The fastest way to run Qwen 3.8 Flash Next and Qwen 3.8 27B on a Mac: 125 tok/s in OpenCode on an M5 Max. Native MTP speculative decoding on Apple Silicon, exact at any temperature. OpenAI and Anth.... It is ranked #321 on the AI Radar, in Open & local models, first seen 2 days ago and shared in 1 post (5.9k views).
GitHub - youssofal/MTPLX: The fastest way to run Qwen 3.8 Flash Next and Qwen 3.8 27B on a Mac: 125 tok/s in OpenCode on an M5 Max. Native MTP speculative decoding on Apple Silicon, exact at any temperature. OpenAI and Anthropic compatible local server. · GitHub Skip to content Navigation Menu Sign in Appearance settings Search / Sign in Sign up Appearance settings You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert {{ message }} youssofal / MTPLX Public…
What people said about mtplx on X
🛸 Mac 圈的外挂来了，MTPLX 能让本地跑的大模型出字速度快到两倍多。 GitHub 2300+ star，开发者 Youssof Altoukhi 实测在 16GB 的 M4 Mac mini 上提速 1.6 倍，M5 Max 上到 2.24 倍。 在自己电脑上跑模型最劝退的就是慢，问一句话，看着字一个一个往外蹦，等得人直接切回网页版。MTPLX 的思路是把模型自带的多 token 预测头用起来：一次先草拟出好几个 token，再用一次批量前向把这一整块验掉，靠精确拒绝采样和残差校正决定留哪些，全程不用另外挂一个小模型来打草稿。 提速不是拿质量换的——项目专门声明在 temperature 0.6、top_p 0.95…
— @Ryrenz, 2 days ago · 54 likes · see the post
Alternatives to mtplx
- Pirate Face — The un-bannable, checksum-verified mirror for open models. Claim your handle before a squatter does.
- atomic.chat — Run AI models locally ->
- deepseek-v4.1-flash-exl3-2x-dgx-sparks — DeepSeek v4.1 Flash EXL3 2.9 bpw for 2x DGX Sparks - MiaAI-Lab/DeepSeek-v4.1-Flash-EXL3-2x-DGX-Sparks
- bespoke-nimble-9b — We’re on a journey to advance and democratize artificial intelligence through open source and open science.
- Cactus Compute — It runs on mobiles, wearables, smart home devices, small robots and microcontrollers, with prebuilt engines for macOS,…
- orcabonsai-27b-uncensored — Open source
mtplx in numbers
- Rank on the AI Radar: #321 of 1190
- Shared in 1 post by 1 account: @Ryrenz
- 5.9k views on those posts
- First seen 2 days ago, last shared 2 days ago
- Pricing seen by Jev: open source
- Market: Open & local models
FAQ
What is mtplx?
The fastest way to run Qwen 3.8 Flash Next and Qwen 3.8 27B on a Mac: 125 tok/s in OpenCode on an M5 Max. Native MTP speculative decoding on Apple Silicon, exact at any temperature. OpenAI and Anth... It was first shared on X 2 days ago and is ranked #321 on the AI Radar.
Is mtplx free?
It is open source.
Who shared mtplx?
1 account on X, including @Ryrenz, in 1 post totalling 5.9k views.
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 27.2k posts from 4.8k X accounts over the last 14 days, 1.2k tools, 19 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-19 18:08 UTC. Full method.