The M5 Ultra combines two M5 Max into a single four-dice package, with 1.2 TB/s of bandwidth. The M6, which is 2 nm, gains 30% GPU compute for AI.
Apple presented on August 25, 2026 two chips that push local calculation for AI. The M5 Ultra, described by the company as its most powerful chip to date, merges two Max M5s — a first four-dose configuration at Apple. It mounts up to 36 processor cores and 80 graphics cores, with 1.2TB/s of unified memory bandwidth, 50% more than the M3 Ultra. The target is explicit: rendered heavy 3D and artificial intelligence loads.
The M6, he, targets the general public. Burned in 2 nm, it adds two processor cores to the M5 to reach twelve, doubles its neuronal motor to sixteen cores, and offers about 30% additional GPU calculation for AI compared to the previous generation. In concrete terms, this accelerates the processing of queries when a language model runs directly on the machine.
Apple insists on this local execution, presented as a privacy choice: developers can run and refine large models without data leaving the Mac. Both chips equip the Mac Mini and Mac Studio, starting at $899 for the first, with deliveries starting on September 22.
Read the full article on the original source
Go to the source →