Apple is moving its Mac silicon to a new process generation with the M6, its first chip built on a 2nm process. The new chip arrives in the Mac mini and brings changes across CPU, GPU, and AI performance compared with the 3nm-based M4.
The M6 uses a 12-core CPU and 12-core GPU, while its Neural Engine configuration has expanded to two 16-core blocks. Apple also claims up to 170GB/s of unified memory bandwidth, giving the chip more headroom for demanding local workloads.
A 2nm shift with more CPU and GPU cores
The M6 adds two CPU cores over the M4, with Apple describing one of its CPU cores as the world’s fastest. The GPU also grows by two cores, while every GPU core now includes a Neural Accelerator.
That combination is aimed at more than traditional Mac performance. Apple claims the M6 can deliver up to 2× faster graphics and up to 4× faster AI performance than the M4 in the Mac mini.
The figures are Apple’s own comparisons, so real-world gains will vary by workload. Applications that can use the additional GPU resources and dedicated AI hardware should have more room to scale than software limited by CPU performance alone.
Neural Engine gets a second block
AI processing is another major focus of the M6. The chip uses a Dual 16-core Neural Engine, which Apple says provides up to 2× the performance of the previous generation.
Apple is positioning that hardware around local AI workloads, including large language models, image generation and AI agents. Keeping more of those tasks on-device can reduce reliance on cloud processing while also making performance less dependent on an internet connection.
The M6 Mac mini starts with 16GB of unified memory and can be configured with up to 32GB. Its 170GB/s memory bandwidth also matters for AI workloads, where moving large amounts of data between compute units can become a limiting factor.
What 2nm means for the M6
The move from 3nm to 2nm is the biggest architectural change behind the M6 generation. A smaller process node gives Apple more transistor density and creates additional room to balance performance, power consumption, and dedicated acceleration hardware.
The M6 is therefore not simply an M4 with extra cores. Apple has paired the process transition with a larger CPU and GPU configuration and a broader set of AI-focused accelerators.
For Mac mini buyers, the more interesting question may be how much of that silicon advantage translates into everyday applications. Developers will determine that over time as more software takes advantage of the M6’s Neural Accelerators and expanded AI hardware.
Apple’s decision to introduce its first 2nm Mac chip in the compact Mac mini also sets an early benchmark for the rest of the M6 family. How aggressively the company scales the same architecture into higher-end Macs could reveal where the new process node matters most.
