MyPrivateClaw

MLX on Apple Silicon: Run Gemma 4 Locally at Full Speed

Apple's MLX framework is 3–4x faster than Ollama for local inference on M series Macs. Here's how to use it.

Guide overview

MLX is Apple's machine learning framework, purpose built for the unified memory architecture of M series chips. For Apple Silicon users, it delivers 3–4x faster inference than Ollama or llama.cpp on the same hardware.