MLX

Apple’s MLX Taps M5 Neural Accelerators for Up to 4x Faster Local LLMs

Apple says macOS 26.2 lets MLX use the M5’s GPU TensorOps for on‑device inference gains.