Apple introduces the M6 and M5 Ultra chips, showcasing a significant increase in AI performance.

      Apple has unveiled the M6 chip in a new Mac mini and the M5 Ultra in the updated Mac Studio, marking its entry into 2nm technology and the debut of its first quad-die M-series design, respectively. The M5 Ultra allows for up to 512GB of unified memory with a bandwidth of 1.2TB/s.

      The question surrounding the Mac mini has been resolved. TNW reported earlier that it was uncertain whether the device would feature an M5 or an M6 chip, and Apple clarified this by introducing the M6, its first 2-nanometer chip.

      The specifications indicate an upgrade. It boasts a 12-core CPU composed of two super cores, four performance cores, and six efficiency cores, a 12-core GPU, and what Apple refers to as a Dual 16-core Neural Engine.

      Apple's claims are significant, stating that the M6 offers the fastest single-threaded performance globally, around 30% greater peak GPU compute for AI compared to the M5, and up to 170GB/s memory bandwidth.

      Details of the launch were highlighted in a report from June, where TNW noted that Apple would showcase a base M6 without Pro or Max versions, utilizing an M5 Ultra as an interim solution in the new Mac Studio, which has turned out to be accurate.

      The Ultra model is particularly noteworthy, as it represents the first quad-die M-series component, connecting two dual-die M5 Max chips via UltraFusion at over 4.4TB/s between them.

      Its specifications include 36 CPU cores and 80 GPU cores, with the capacity for 512GB of unified memory at 1.2TB/s, which Apple claims is significantly higher bandwidth than the M3 Ultra.

      One key point from the announcement stands out: Apple asserts that the machine is capable of running models with hundreds of billions of parameters entirely on-device.

      The memory capacity presents a challenge in a market currently facing a DRAM shortage, which has led to empty Mac shelves and rising prices across the industry.

      For European consumers, the implications are more legal than technical. A locally running model ensures that prompts, documents, and customer data remain on-site.

      This addresses a concern highlighted by TNW, which revolves around the political aspect of European organizations processing sensitive information on infrastructure beyond their control.

      However, it is important to note that sovereignty is not the main issue, as the chip is designed by an American company and manufactured in Taiwan. Therefore, local inference is determined by the location of the data, not by the provider of the computing power.

      Nonetheless, for entities like hospitals, law firms, or government agencies that are restricted from placing their files in external clouds, a desktop-sized unit capable of handling extensive models is among the most practical offerings Apple has introduced in years.