Apple Launches M6 and M5 Ultra Macs for Local AI Inference
The unified memory architecture, 2nm process, and high memory bandwidth make these machines capable of running large AI models locally without cloud infrastructure. The M5 Ultra's 512GB memory ceiling and multi-system clustering extend that to very large models.
Apple has launched the Mac mini with the M6 chip and the Mac Studio with the M5 Ultra, both targeting local AI inference and software development. The Mac mini M6 delivers on-device AI performance up to 4x faster than its predecessor. The M6 is Apple's first 2nm chip in the M-series Mac lineup, featuring a 12-core CPU, 12-core GPU, dual 16-core Neural Engine, and 170GB/s memory bandwidth. The M5 Ultra joins two dual-die M5 Max chips via UltraFusion into Apple's first quad-die SoC, with up to 36 CPU cores, 80 GPU cores, 512GB unified memory, and 1.2TB/s memory bandwidth, 50% more than M3 Ultra. Mac Studio with M5 Ultra ships configurable up to 256GB unified memory now, with a 512GB configuration arriving in October. Apple claims the M5 Ultra can hold LLMs with hundreds of billions of parameters entirely on-device. Apple also supports Thunderbolt 5 clustering with RDMA, allowing multiple Mac Studio systems to pool memory for distributed inference.
The Neuron Daily reported August 26 that the Mac mini M6 delivers on-device AI performance up to 4x faster than its predecessor, a concrete figure absent from Apple's initial coverage of the August 25 announcement introducing the M6 as the company's first 2nm chip in its Mac lineup.
Apple announced the Mac mini with M6 and Mac Studio with M5 Ultra on August 25 for local AI inference.
The M6 is Apple's first 2nm chip in the M-series Mac lineup; the M5 Ultra joins two M5 Max dies via UltraFusion into Apple's first quad-die SoC, with up to 512GB unified memory and 1.2TB/s bandwidth, 50% more than M3 Ultra. Apple says the M5 Ultra can hold LLMs with hundreds of billions of parameters on-device; the 512GB configuration ships in October.
Apple announced Mac mini with M6 and Mac Studio with M5 Ultra, targeting local AI inference.
I send a short email each morning with the stories that moved. If you would rather just read here, that works too.
Subscribe free