Apple's new M5 MacBook lineup went on sale today, March 11, across both the MacBook Air and MacBook Pro lines — marking the first time Apple has shipped Neural Accelerators inside every GPU core on a Mac.

What's New in M5

The base M5 chip, powering the MacBook Air, delivers roughly 15% better multi-core CPU performance over the M4 generation (Geekbench 6: 17,073 vs 14,731). More notable is the GPU redesign: each core now includes a dedicated Neural Accelerator, enabling hardware-accelerated AI tasks that previously required software emulation. In Apple's own testing, this translates to up to 6.9× faster AI video enhancement over the M1 generation in Topaz Video.

The MacBook Air also ships with doubled base storage at 512GB, Apple's N1 wireless chip enabling Wi-Fi 7 and Bluetooth 6, and up to 18 hours of battery life.

M5 Pro and M5 Max: A Bigger Leap

The MacBook Pro variants carry M5 Pro and M5 Max chips built on Apple's new Fusion Architecture — a two-die SoC design that scales CPU and GPU resources while preserving the unified memory model. Apple claims over 4× peak GPU compute for AI versus M4 Pro and M4 Max, driven by Neural Accelerators baked into each of the up-to-40 GPU cores. The M5 Max reaches a Geekbench 6 multi-core score of 29,233.

Both chips include Memory Integrity Enforcement — Apple's first always-on hardware memory safety feature, designed to block memory corruption attacks without performance overhead. MacBook Pro also gains Thunderbolt 5 support for the first time.

Why It Matters for AI

The local AI inference market has been growing steadily, with developers running open-source LLMs on-device to cut cloud costs and latency. The M5 Pro/Max chips represent a meaningful step toward consumer-grade hardware capable of running larger models at practical speeds. Apple positions the MacBook Pro as the primary target for developers building and running AI workloads locally.