Apple debuts M6 and M5 Ultra silicon built for local AI

Apple just announced the M6 and M5 Ultra chips, shifting its desktop silicon hard toward running heavy AI models on your desk.

Apple introduced the M6 inside a new Mac mini and the M5 Ultra inside a new Mac Studio. The M6 is Apple’s first 2-nanometer chip, featuring a 12-core CPU, a 12-core GPU with Neural Accelerators, a Dual 16-core Neural Engine, and up to 170GB/s of memory bandwidth. Meanwhile, the M5 Ultra debuts Apple's first quad-die architecture, packing up to a 36-core CPU, an 80-core GPU, and 1.2TB/s of unified memory bandwidth—50 percent more bandwidth than the M3 Ultra.

Why it matters: Running large AI models locally requires massive memory throughput and dedicated silicon. By combining quad-die design with Neural Accelerators across the GPU, Apple is making it easier for developers to run massive AI models right on a desktop without renting cloud servers.

Know this: Apple Intelligence features designed for the new hardware will arrive with macOS 27 this fall, with public testing currently open in the Apple Beta Software Program.

If you want to run massive local models without building a noisy server rack, Apple just built the box for it.