Synor

Research: vllm-mlx on Apple Silicon achieves 21% to 87% higher throughput than llama.cpp

Posted by Synor@reddit | LocalLLaMA | View on Reddit | 23 comments