Apple introduces M5 Ultra and M6 chips for new Mac systems
Apple has announced the M5 Ultra and M6 processors for its new Mac Studio and Mac mini models. The M5 Ultra, described by Apple as its most powerful chip yet, is the company’s first quad-die chip and offers an up-to-36-core CPU, an up-to-80-core GPU and 1.2 TB/s of unified memory bandwidth.
Apple says the M5 Ultra is formed by fusing two dual-die M5 Max chips. Its memory bandwidth is 50% higher than that of the M3 Ultra, which Apple released about a year and a half ago. The company positions the new processor for compute-intensive professional work, including complex 3D rendering, visual effects and running frontier AI models.
M5 Ultra targets demanding local AI workloads
The Mac Studio configuration with M5 Ultra is aimed at professionals, but it may also interest hobbyist AI developers. Apple’s emphasis is on workloads that can be run locally, combining CPU, GPU and unified memory resources in a single system.
That local approach is central to Apple’s developer pitch. Apple says its tools and frameworks automatically optimise performance across the CPU, GPU and Neural Engine. Developers can use Apple Foundation Models, App Intents for Apple Intelligence capabilities, or proprietary models to build and run AI workloads entirely on device.
M6 adds CPU cores and AI compute
The M6 is designed more for everyday users, Apple said. Sri Santhanam, Apple vice president of Silicon Engineering, said the chip is built on a 2 nm process and combines a new CPU complex, two additional CPU and GPU cores, a Dual 16-core Neural Engine and increased unified memory bandwidth with an emphasis on energy efficiency.
Its 12-core CPU complex adds two cores compared with M5. Apple says M6 can provide nearly a 30% increase in peak GPU compute for AI over M5, improving prompt-processing speed when on-device large language models are used.
Mac availability and developer choices
Apple’s newest Mac mini models are available for pre-order from $899 and will ship after September 22. The announcements arrive as Apple expands the ways software can work with AI models, including Apple opening AI model choice in iOS 27 and its implications for developers building applications around local and external model capabilities.
For businesses, the practical implication is to assess which AI tasks require local execution and whether the M5 Ultra’s capacity or the M6’s efficiency better fits the intended Mac deployment.