Apple unleashes M5, the next big leap in AI performance for Apple Silicon (www.apple.com)

Apple today unveiled the M5 system-on-chip, a major generational jump for Apple silicon focused on AI and graphics. Built on third‑generation 3nm, M5 adds a next‑generation 10‑core GPU with a dedicated Neural Accelerator in each core, delivering over 4× peak GPU compute for AI versus M4 (and >6× vs M1). It pairs a up-to‑10‑core CPU (six efficiency + up to four performance cores) with Apple’s “fastest” performance core, a faster 16‑core Neural Engine, a more powerful media engine, and 153 GB/s unified memory bandwidth (≈30% higher than M4) with up to 32 GB capacity. Graphics gains include third‑generation ray tracing and up to 45% uplift in ray‑traced workloads. M5 will ship in the 14‑inch MacBook Pro, iPad Pro and Apple Vision Pro. For AI/ML users and developers the M5’s biggest implications are practical: much larger and faster on‑device models and real‑time workloads (local LLMs, diffusion/image generation, AR/VR inference) without constant cloud roundtrips, lower latency and improved privacy. Apple says existing frameworks (Core ML, Metal Performance Shaders, Metal 4) benefit automatically, while Metal’s Tensor APIs let devs target the new Neural Accelerators directly. Higher unified bandwidth and integrated accelerators should enable more efficient multi‑component pipelines (GPU + Neural Engine + CPU), faster Apple Intelligence features, and better XR visuals and refresh rates — all with Apple’s emphasis on power efficiency for reduced lifetime energy impact.