Ollama Now Runs Faster on Macs Thanks to Apple's MLX Framework

Ollama updated its Mac app to use Apple’s MLX framework, boosting on-device performance.

The company reports about 1.6× faster prefill speed and nearly double decode speed on Apple Silicon Macs.

M5-series Macs see the largest gains and the update adds smarter memory management for longer sessions.

The preview release is Ollama 0.19, requires over 32GB unified memory, and currently supports Alibaba's Qwen3.5.