When it comes to the 2026 model of the 16-inch MacBook Pro, the tiny chips inside have the biggest changes.
For starters, in the case of the M5 Max, the CPU offers more cores, increasing from 15 cores in the M4 Max to 18 in the M5 Max, effectively giving the MacBook Pro a bigger engine for multitasking. This is in addition to the dedicated 16-core Neural Engine and increased memory bandwidth, which is boosted to 614GBps, up from the 546GBps of the M4 Max.
(Credit: Apple)
The cores themselves are also upgraded, with Apple introducing a new category of core with the M5 Pro and M5 Max, dubbed the “super core.” This new third kind of core is optimized for single-threaded performance, offering the fastest single-core speeds of any Apple Silicon chip. Also, for the higher-powered M5 Pro and Max chips, Apple has dropped the lowest-power “efficiency” cores and designed the CPU solely around super cores and traditional performance cores. (Meanwhile, the vanilla M5 uses just super cores and efficiency cores.) In a sense, with M5 Pro and Max, Apple is taking its usual “big.LITTLE” core hierarchy and, in these chips, adopting something you might call “bigger.BIG,” with traditional performance cores as the lower-tier cores. These performance cores are the multithread workhorses, offering better efficiency and dynamic range than the super cores, but without the limitations that sometimes made efficiency cores (the third tier of chip core) into a bottleneck.
New cores are just part of the puzzle, though. Apple has made great strides in combining the CPU with high-performance GPU cores, often going toe-to-toe with Nvidia in laptop graphics performance while maintaining a unified design that shares memory and resources for speedy work. But as the GPU half of that chip grows, it becomes that much harder to make. So the M5 Pro and M5 Max are actually getting a pair of chip dies, unified by Apple’s new Fusion interconnect, to function as one.
This approach lets Apple produce smaller, specialized dies for the CPU and GPU, improving the yield of usable chips. The company can then mix and match dies to package only the good ones. The Fusion technology has its roots in the UltraFusion interconnect used to fuse two full CPU chips in the M-series Ultra lineup, but the purpose here is probably to make the chip fabrication process a bit more flexible.
The M5 Max’s 40 GPUs also get an upgrade with a Neural Accelerator added to each graphics core. This change was introduced on the base M5 chip, and uses technology from the AI-focused Neural Engine, adding it to each GPU core to boost both graphics rendering and AI performance for machine learning and other AI workloads. It was impressive enough with just 10 GPU cores on the M5, but it’s a dramatic improvement with 40 GPU cores on the M5 Max.
