Running local models on Macs gets faster with Ollama's MLX support a arstechnica.com 2 kinoriw 7 months ago 0 comments