Run the Apple Silicon MLX build
run with ollama
// what it does
The -mlx tags (gemma4:e2b-mlx, gemma4:e4b-mlx, gemma4:12b-mlx, gemma4:26b-mlx, gemma4:31b-mlx) use Apple's MLX runtime for faster, more memory-efficient inference on M-series Macs than the default GGUF path. Download sizes are comparable to the plain tags.
// shell
$ ollama run gemma4:e4b-mlx// gotcha
MLX tags only run on Apple Silicon. Pulling gemma4:e4b-mlx on an x86 or NVIDIA box wastes the download because the runtime cannot load it — use the plain gemma4:e4b there.