Apple has officially released its new M6 and M5 Ultra chips. M6 is Apple’s first processor built using a 2nm process and is the first to be installed on the new Mac mini. On the other hand, M5 Ultra further pushes the specifications of desktop Macs to new heights, with a maximum configuration of 36-core CPU, 80-core GPU, and 512GB of unified memory. It is mainly targeted at AI, 3D rendering and professional workflows.
M6 uses a new 12-core CPU, which is composed of 2 super cores, 4 performance cores and 6 energy-saving cores. Compared with M5, there are two more CPU cores. Apple claims that the multi-threading performance is up to 1.2 times that of the M5, and is equipped with a 12-core GPU and dual 16-core neural network engines.
The GPU part also adds two more cores than the previous generation, and adds a neural network accelerator to each GPU core. M6 supports up to 32GB of unified memory, and the memory bandwidth is increased to 170GB/s. Apple said that the GPU AI peak computing power is nearly 30% higher than that of M5.
The new Mac mini also becomes the first product to be equipped with M6. Apple said that compared with the Mac mini equipped with M4, the new machine’s CPU performance is up to about 40% higher, graphics performance is up to 2 times, and AI workload is up to 4 times higher. The connectivity capabilities are simultaneously upgraded to Wi-Fi 7, Bluetooth 6 and standard 2.5Gb Ethernet, with optional 10Gb Ethernet.
The Taiwan market has opened pre-orders on August 27. The Mac mini equipped with M6 starts at NT$29,900, and the M5 Pro version starts at NT$59,900. It is expected to be available from September 22.
If the M6 was Apple’s first step in bringing 2nm to mainstream desktop Macs, the M5 Ultra is a direct attack on the workstation and native AI markets.
M5 Ultra adopts the new generation of UltraFusion technology, forming a four-die architecture for the first time in the M series SoC, providing up to 36-core CPU and 80-core GPU. Each core of the GPU also adds a neural network accelerator. Apple claims that the peak performance of its AI GPU can be up to 4.5 times that of the M3 Ultra, and the graphics processing performance can be improved by up to about 40%.
The most surprising thing is the memory specifications. M5 Ultra can be configured with up to 512GB of unified memory, and the bandwidth reaches 1.2TB/s, which is 50% higher than M3 Ultra. Such a large unified memory capacity allows the system to load large data sets directly locally and even execute large language models with hundreds of billions of parameters.
The new Mac Studio will be available in two versions: M5 Max and M5 Ultra. The M5 Max starts at NT$84,900 and the M5 Ultra starts at NT$199,900. Taiwan has also opened pre-orders on August 27 and will start shipping on September 22. The unified memory version up to 512GB is expected to be launched at the end of October.
In addition to performance, the new Mac Studio also adds Wi-Fi 7, Bluetooth 6, and Thunderbolt 5 up to 120Gb/s, and supports up to eight displays, or four full 5K, 120Hz Studio Display XDR.
From the M6’s introduction of 2nm for the first time to the M5 Ultra’s increased unified memory to 512GB, Apple is clearly pushing its Mac desktop products further into the AI and professional workstation markets.
For ordinary users, the M6 Mac mini is still the most accessible new product in this update; while the M5 Ultra Mac Studio has clearly targeted large-scale native AI, 3D, video production and scientific research workloads, and its specifications have gradually opened up a different product line from traditional high-end workstations.
