youssofal/MTPLX
Apple Silicon LLM inference engine
MTPLX is a native Mac application and command-line tool that accelerates local language model inference on Apple Silicon using multi-token prediction. It provides an OpenAI and Anthropic compatible server for running specific Qwen and Bonsai models with high throughput.
- Stars
- 2,400
- Stars gained in 7 days
- +100
- Contributors
- 44
- Forks
- 189
- Days trending in the last 30
- 8 days