Apple launches its first 2nm M6 chip and quad-die M5 Ultra

Apple introduced the M6 and M5 Ultra processors on Tuesday. The M6 debuts inside a redesigned Mac mini as the company’s first processor built on a 2nm process, while the M5 Ultra powers a refreshed Mac Studio and is described by Apple as its most powerful chip ever.

The M6 becomes Apple's first 2-nanometer processor. The company reports that the node delivers higher transistor density together with better power efficiency. Those gains reportedly let engineers expand compute resources without enlarging the die.

A 12-core CPU sits at the heart of the M6, two cores more than the M5. The design marks Apple’s first processor to combine all three core types: two super cores focused on peak single-threaded speed, four performance cores tuned for intensive multithreaded workloads, and six efficiency cores that manage lighter background tasks. Apple claims the configuration achieves “the world’s fastest single-threaded performance” and delivers “up to 1.2x faster multithreaded performance as compared to M5.”

M6 GPU and memory see targeted upgrades. The GPU scales to 12 cores, each equipped with a dedicated Neural Accelerator, producing a 30 percent rise in peak GPU compute for AI workloads versus the M5. Additional refinements target the shader core architecture, Dynamic Caching, and hardware-accelerated ray tracing. Memory support reaches 32 GB of unified RAM at up to 170 GB/s bandwidth, an improvement Apple pegs at 10 percent over the M5. The chip also adds native hardware processing of the FP8 data format, an 8-bit standard intended to accelerate AI tasks while reducing memory demands.

Benchmarks position M6 ahead of prior generations. Official figures show multithreaded CPU performance climbing as much as 1.2x over the M5 and 2.4x over the M1. Peak GPU compute for AI is nearly 30 percent higher than the M5 and more than 8x that of the M1. Geometry throughput for complex graphics rises 50 percent, while the dual 16-core Neural Engine supplies up to 2x the peak compute of earlier generations. Memory bandwidth hits 170 GB/s, 10 percent above the M5 and 2.5x the M1.

M5 Ultra adopts a quad-die architecture. The new flagship links two dual-die M5 Max chips through next-generation UltraFusion interconnect technology. Apple states the revised link supplies more than 4.4 TB/s of inter-die bandwidth and over 6x higher connection density, enabling all four dies to function as one unified processor.

The M5 Ultra CPU scales to 36 cores comprising 12 super cores and 24 performance cores. Relative to the M3 Ultra, Apple cites gains of up to 1.25x in single-threaded performance and 1.3x in multithreaded performance. The company again calls the M5 Ultra its “most powerful chip ever.”

M5 Ultra GPU and memory scale significantly higher. Graphics reach up to 80 cores, each carrying a Neural Accelerator. Apple projects up to 4.5x higher peak GPU compute for AI versus the M3 Ultra and as much as 40 percent faster overall graphics performance. The GPU incorporates the latest shader core with second-generation dynamic caching, hardware-accelerated mesh shading, and third-generation ray tracing. Unified memory tops out at 512 GB with 1.2 TB/s of bandwidth, a 50 percent increase over the M3 Ultra. GPU AI compute is listed at more than 6x the level of the M1 Ultra.

Both chips are available immediately in their respective new desktop systems.