Ahead of the iPhone 18 Pro event announcement, Apple surprised customers by introducing not only its most powerful Mac processor yet, the M5 Ultra, but also the next generation of the M6 chip, with a base version for the Mac mini. These two processors are unique in their own way, with the M6 variant found in the Mac mini and the M5 Ultra being available exclusively on Mac Studio. Still, they focus a lot on AI and performance improvements, making them not only better than their immediate predecessors but also a huge leap forward from the M1 family.
As Apple continues its cross-generation strategy between its chips – with the previous Mac Studio, the company offered M4 Max and M3 Ultra variants – rumors speak of a short M6 family. Aside from a few other Mac refreshes, Apple might skip the Pro, Max, and Ultra variants altogether, as it’s already working on the M7 models, with an increasing focus on handling AI workloads.
According to Bloomberg, this will lead Apple to introduce the long-rumored MacBook Pro with OLED with the M5 Pro and M5 Max chips, instead of the M6 Pro and M6 Max variants. Still, even if the rumors are true and Apple quickly abandons the M6 generation, there is a lot to hope for, as this processor is Apple’s first chip with 2 nanometers, making it more powerful and efficient.
Getting to know Apple’s M6 chip
Apple says the M6 processor, the first built with 2nm process technology, is perfect for the workflows of students, developers and everyday users because it combines excellent performance with on-device AI. With an all-new 12-core processor, offering two more cores than its M5 counterpart, it includes two super cores, four performance cores and six efficiency cores. The company claims that single-threaded performance is up to 1.2x faster than the M5, and up to 2.4x faster than the M1, which is perfect for very demanding tasks, while handling everything else in the background smoothly.
The M6 GPU also has 12 cores, still two more than the previous generation, with Neural Accelerator in each core. While the jump between M4 and M5 was already impressive, Apple’s new chip may be nearly 30% better in terms of GPU computing for AI than the M5. The company also says it’s improving core shader architecture, dynamic caching, and hardware-accelerated ray tracing, although they haven’t been rebranded next-gen. Still, the company expects everything to run much smoother, including games like Cyberpunk 2077.
For the M6 chip, Apple supports up to 32GB of unified RAM, with up to 170GB/s of unified memory bandwidth, which is a 10% improvement over the M5, but a 2.5x increase over the first Apple Silicon. For everyday customers, this means that even more data can be read at the same time.
Inside Apple’s most powerful processor yet
Exclusive to Mac Studio, the M5 Ultra is Apple’s most powerful processor yet. Although it still uses a 3nm manufacturing process, it takes advantage of UltraFusion, an Apple technology that connects two M5 Max chips together to deliver up to 4.4 TB/s of bandwidth and incredibly high specifications for the most demanding users. According to Apple, the M5 Ultra can get up to 36 CPU cores, with 12 super cores and 24 performance cores, making single-threaded performance 1.25 times faster and multi-threaded performance 1.3 times faster than the latest top chip in the Apple M3 Ultra – and the 32-core Neural Engine.
With a massive 80-core GPU, the largest on an Apple Silicon Mac to date, it includes neural accelerators for each, making the cutting-edge GPU 4.5 times faster than the M3 Ultra, which still lacked this technology. This also means that this chip provides second-generation dynamic caching, hardware-accelerated mesh shading, and third-generation ray tracing, making graphics performance 40% faster, a huge benefit for the most demanding users.
The M5 Ultra also has up to 512GB of unified RAM, which can only be ordered in October. That said, at the moment users can only pre-order an additional $4,000 for 256GB unified memory. However, once this option becomes available, users will have 50% more memory bandwidth to run huge LLMs on this device. In addition, the processor also has dedicated H.264, HEVC and ProRes hardware encoding and decoding engines, as well as other solutions for high-end video editing.
