Apple has today announced its new M5 processor, the latest addition to its custom silicon lineup focused on AI performance.
The M5 features dedicated Neural Accelerators embedded directly into each of the GPU’s ten cores, delivering over 4x the AI compute performance compared to the M4.
Beyond the AI-focused GPU architecture, the M5 brings meaningful improvements across the board. The CPU delivers 15% faster multithreaded performance, while graphics see up to a 45% boost in ray-traced applications. Memory bandwidth jumps nearly 30% to 153GB/s (gigabytes per second), and the 16-core Neural Engine gets faster as well.
The chip is manufactured on third-generation 3nm technology and supports up to 32GB of unified memory.
Where You’ll Find It
The M5 launches in three products, all available for pre-order today:
- 14-inch MacBook Pro
- iPad Pro
- Apple Vision Pro
For Vision Pro specifically, the M5 enables more pixel rendering and higher refresh rates that should translate to noticeably sharper visuals and smoother performance.
As Apple continues rolling out its Apple Intelligence features, the M5’s architecture allows these AI tools to run faster entirely on-device. Larger models can now run locally thanks to the increased memory bandwidth and capacity, aligning with Apple’s privacy-first approach.
The embedded Neural Accelerators also mean developers using Apple’s frameworks (Core ML, Metal Performance Shaders, and the new Metal 4) will see automatic performance gains in AI-driven applications.
By making AI acceleration a core function rather than a bolt-on feature, the company is betting that machine learning workloads will be as routine as opening a web browser. For users running image generation tools, large language models, or Apple’s own Intelligence features, that bet translates to meaningfully faster performance.
For complete technical specifications and additional details, see Apple’s official release post.
Leave a Reply