Meanwhile, Apple has made good on some of its M5 chip speed claims that were previously hard to verify because they relied on pre-release software. Apple’s open-source MLX frameworks now support the M5’s Neural Accelerator, which enables dramatic speed improvements.

Federico Viticci of MacStories was also able to test the pre-release MLX frameworks on an M5 iPad Pro and reports some pretty spectacular results. Obviously, these are the earliest days, and it’s a prerelease framework, but it seems like the 26.2 updates will unlock a lot of the promised horsepower of the M5—at least on the AI-related GPU front.

Another Apple update in 26.2 will enable even higher-speed Mac performance in clusters of Macs. This new feature uses the 480Gb/s of the Mac Studio’s six Thunderbolt 5 ports to run shared workloads across (for example) multiple M3 Ultra Mac Studios, creating AI clusters that have enormous processing power while using potentially an order of magnitude less electricity than a more traditional processing cluster.

I have no idea how practical that is, but it’s clear that Apple sees a place where high-end Apple silicon Macs can play in the AI space, and it’s moving further in that direction with these 26.2 updates.

All of this stuff is in beta right now, but you should expect 26.2 final to ship before the end of the year.

If you appreciate articles like this one, support us by becoming a Six Colors subscriber. Subscribers get access to an exclusive podcast, members-only stories, and a special community.