Apple Newsroom:

Apple today announced M5, delivering the next big leap in AI performance and advances to nearly every aspect of the chip. Built using third-generation 3-nanometer technology, M5 introduces a next-generation 10-core GPU architecture with a Neural Accelerator in each core, enabling GPU-based AI workloads to run dramatically faster, with over 4x the peak GPU compute performance compared to M4. The GPU also offers enhanced graphics capabilities and third-generation ray tracing that combined deliver a graphics performance that is up to 45 percent higher than M4. M5 features the world’s fastest performance core, with up to a 10-core CPU made up of six efficiency cores and up to four performance cores. Together, they deliver up to 15 percent faster multithreaded performance over M4. M5 also features an improved 16-core Neural Engine, a powerful media engine, and a nearly 30 percent increase in unified memory bandwidth to 153GB/s. M5 brings its industry-leading power-efficient performance to the new 14-inch MacBook Pro, iPad Pro, and Apple Vision Pro, allowing each device to excel in its own way. All are available for pre-order today.

The M5 14-inch MacBook Pro is not accompanied by its more powerful siblings, which feature an extra USB Type-C port on the right and the Pro and Max chip variants. Those are reportedly delayed until January 2026, just to be replaced by redesigned models with organic-LED displays later in the year. I’ve been on the record as saying the base-model MacBook Pro is not a good value, and I mostly share the sentiment this year. The M5 has better graphics cores and an improved Neural Engine, both for on-device artificial intelligence processing. Third-party on-device large language model apps typically use the graphics processing unit to run the models, whereas Apple Intelligence, being optimized for Apple silicon, uses the Neural Engine. On the Mac, these updates are insignificant for now because the M4 Pro and M4 Max, which Apple still sells, have better GPUs than the M5. But on the iPad Pro, where the only comparison is the M4, on-device LLMs run at their fastest yet.

This more or less matches the framing Apple’s marketing seems to imply. The M5 MacBook Pro is centered around better battery life and marginally improved performance across the board compared to older generations like the M1 and M2, whereas the iPad Pro is positioned as an on-device AI powerhouse. The rationale is simple: There are more powerful Macs to run LLMs on for sale today, but there aren’t more powerful iPads. That will, of course, change come next year when the M5 Pro, M5 Max, and later the M6 generation are announced, but for now, the M5 MacBook Pro is middle of the road. I’d tell all prospective M5 MacBook Pro buyers to wait three months and spend an extra $400 for the M5 Pro version, or, better yet, wait a year for the redesigned M6 Pro MacBook Pro. (Sent from my M3 Max MacBook Pro I was planning on upgrading this year, had Apple not staggered the releases.)

The story of the iPad Pro is nothing revolutionary. It only has one front-facing camera, contrary to what Mark Gurman, Bloomberg’s Apple reporter who’s typically correct about almost every leak, said. It does, however, ship with the N1 Wi-Fi 7 and Bluetooth 6 processor, along with the C1X cellular modem on models that need it. The base storage configurations also have more unified memory for on-device LLMs — 12 gigabytes — and the prices remain the same. Coupled with iPadOS 26 improvements, the iPad Pro is probably the highlight of Wednesday’s announcements purely because they enable much larger, power-hungry LLMs to run on-device. While this is probably insignificant for the low-quality Apple Intelligence foundation models that run perfectly fast on even older A-series processors, it is important to use more performant LLMs like GPT-OSS, my favorite so far.

And then there’s Apple Vision Pro, perhaps the most depressingly hilarious announcement on Wednesday. The hardware, with the sole exception of the M5 (upgraded from the M2), is entirely untouched. Apple touts “10 percent more pixels rendered” due to the enhanced processor, but that’s misleading: The M5 only decreases visionOS’ reliance on foveated rendering, the technique that allows the headset to only render what a user is actively looking at to conserve resources. The display panels are the exact same, down to every last pixel, but the device now renders 10 percent more pixels, even when a user isn’t looking directly at them. These pixels will only be visible in a user’s peripheral vision. Rendered (not passthrough) elements are also displayed at 120 hertz instead of 90 hertz, but the difference is imperceptible to me when comparing my various ProMotion devices to Apple Vision Pro. (It’s a meaningful difference in terminology that Apple didn’t call Apple Vision Pro’s displays “ProMotion” anywhere, because they’re not.)

A new band ships with the headset by default: It is now two individually adjustable Solo Knit Bands conjoined. One is placed at the back of the head, similar to the Solo Knit Band that shipped with the original Apple Vision Pro, while the other sits at the top to provide additional support. I’m sure it’s much more comfortable than either original band — both of which are still available for sale — but I’m not about to spend $100 on a product I haven’t touched since June. For Apple Vision Pro connoisseurs, however, I’m sure it’s a good investment. And of course, nobody with a launch-day device should buy an M5-equipped Apple Vision Pro, especially because there is no trade-in program for the product. Even Apple doesn’t want them back.