youssofal/MTPLX

3x faster speeds on MLX | Qwen 3.8 27B | Native MTP Speculative Decoding On Apple Silicon With No External Drafter.

↗ Open on GitHub

Stars

1,944

Forks

145

Language

Python

License

Apache-2.0

Last push

1 week ago

Python metal qwen apple-silicon mlx inference-engine openai-compatible speculative-decoding local-ai qwen3-next anthropic-compatible mtp mtplx native-mtp speculative-sampling

No related intel yet

This repo has not appeared in any of the sources the radar tracks. The collector runs on a schedule — check back once it covers this repo.