youssofal/MTPLX
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.
- Stars
- 2.7k
- Forks
- 210
- Language
- Python
- License
- Apache-2.0
- Last commit
- updated Oct 2026
Straight from GitHub
Nobody has written up this repo. Everything on this page comes straight from the GitHub API, so there is no founder write-up, no pros and cons, and no difficulty rating — rather than a generated guess at them.
Maintenance
Pushed within three months of the last stats read (as of Nov 6, 2026).
Licence
Apache-2.0
Tech stack
Primary language: Python. GitHub reports only the dominant language, so this is not the whole stack.
Install
No install command has been verified for this repo. Follow the README rather than a guess.
Quality score
Computed by the discovery import when this entry was added.
- Recent activity×0.30
- 1.00
How recently the default branch was pushed to.
- Adoption×0.20
- 0.73
Stars on a log scale, so adoption counts without dominating.
- Description and topics×0.15
- 1.00
Whether the repo describes itself: description, topics, homepage.
- README present×0.15
- 1.00
Whether the repository has a README at all.
- Licence clarity×0.10
- 1.00
Whether GitHub can resolve the licence to a recognised SPDX id.
- Issue health×0.10
- 1.00
Open issues relative to adoption, as a rough maintenance signal.
Similar repos
Same category, ranked by shared topics.
MLX-VLM is a package for inference and fine-tuning of Vision Language Models (VLMs) on your Mac using MLX.
5.8k starsPythonMITupdated Nov 2026AI & MLLLM inference server with continuous batching & SSD caching for Apple Silicon — managed from the macOS menu bar
23.4k starsPythonApache-2.0updated Nov 2026AI & MLSemantic grep powered by Jina embeddings v5 (MLX on Apple Silicon)
248 starsPythonApache-2.0updated May 2026AI & MLHundreds of models & providers. One command to find what runs on your hardware.
38.7k starsRustMITupdated Nov 2026AI & ML
Stats updated Nov 6, 2026. Stars, forks, issue counts, and dates are a snapshot read from the GitHub REST API, not a live figure. This site makes no requests to GitHub when you load a page.