Why the iPhone 18 Pro's A20 Pro Beat Apple's Own Mac Chip

The headline number from this week's iPhone 18 Pro benchmarks is not that Apple's new phone chip beat desktop processors from Intel and AMD. It's that it beat Apple's own top-of-the-line Mac chip. That's the detail that should make you pause before treating this as just another spec-sheet flex, because it exposes something odd about how "fastest chip" claims get made in 2026, and what they actually predict about the device in your hand.

Key takeaways

The A20 Pro's Geekbench 7 single-core score, clustering around 4,000-4,042 points in early tests, edges out Apple's own M5 Max chip, which scored roughly 3,731 in the same test, according to Digital Trends and Notebookcheck.
Tom's Hardware reports the A20 Pro beats AMD's Ryzen 9 9950X3D by about 26% and Intel's Core i9-14900KS by about 32% in single-core Geekbench 7 scores.
The A20 Pro is Apple's first chip built on TSMC's 2-nanometer (N2) process, with performance cores clocked up to 4.93GHz, per MacRumors and Notebookcheck.
None of this changes the multi-core picture: desktop CPUs and Apple's own M5 Max still beat the A20 Pro by a wide margin on sustained, multithreaded work.

What Actually Happened With the A20 Pro

Apple unveiled the A20 Pro on September 9, 2026, alongside the iPhone 18 Pro and Pro Max, which went on sale September 18, according to MacRumors and multiple outlets tracking the launch. It's a 6-core chip: two new "Super" performance cores plus four efficiency cores, with Apple describing the performance cores as up to 20% faster than the prior generation and calling them, in its own marketing language, "desktop-class" and "the fastest in any smartphone."

Early Geekbench 7 leaks, gathered before the phones were widely in reviewers' hands, put the single-core score around 4,000 to 4,042 points, with a wider spread of roughly 3,927 to 4,725 depending on the specific test run and device, per MacRumors and PC Guide. That variance is normal for pre-release benchmark leaks and will narrow once independent reviewers run their own units. But the headline figure held up across enough outlets to be treated as real: this is now the highest single-core score Geekbench 7 has recorded, on any device.

The Awkward Part Is That It Also Beat Apple's Own Mac Chip

The comparison that got the most attention wasn't against Windows PCs. It was against the M5 Max, the chip inside Apple's current top MacBook Pro. Digital Trends, Notebookcheck and Inven Global all reported that the A20 Pro's single-core score of about 4,042 topped the M5 Max's roughly 3,731 in the same benchmark. That means Apple's phone chip is, by this one measure, faster per core than the chip in its flagship laptop.

That is genuinely strange, and it's worth sitting with why. The M5 Max is built for a laptop with a fan, a battery the size of a paperback, and a chassis designed to shed heat continuously. The A20 Pro is built for a phone that's a few millimeters thick with no active cooling at all. If the phone chip wins on peak single-core speed anyway, the reason has less to do with raw engineering superiority and more to do with how each chip is allowed to spend its thermal budget, which is the real story here.

Person holding a smartphone showing its screen

Why a Phone Chip Can Beat a Desktop CPU on One Number

Geekbench 7's single-core test is short. It measures how fast one core can burst through a fixed set of tasks, not how the chip behaves over ten minutes of sustained load. A20 Pro's performance cores hit up to 4.93GHz, per Notebookcheck, which is only possible because Apple designed them to sprint for a few seconds, not sustain that clock for a video export. Desktop chips from Intel and AMD are tuned differently: they carry far more cores (Ryzen 9 9950X3D has 16, more than double the A20 Pro's core count) built to divide up heavy, sustained workloads, at the cost of not being able to hit quite the same instantaneous per-core peak that a fresh 2-nanometer mobile design can.

The 2-nanometer process matters here too. TSMC's N2 node is a new manufacturing generation, and being first to use it, as Apple is with the A20 Pro, typically buys a real efficiency and clock-speed advantage before rivals catch up on the same node, according to MacRumors and Tom's Hardware's coverage of the launch. Tom's Hardware's own numbers put the A20 Pro's single-core score about 26% ahead of the Ryzen 9 9950X3D and about 32% ahead of the Core i9-14900KS. Those are real, reported figures. They're also measuring one narrow thing extremely well.

Close-up of a circuit board with chip components

Multi-Core Tells a Completely Different Story

Here is the honest counterpoint that most of the coverage buried below the fold: none of this makes an iPhone faster than a workstation, or even faster than Apple's own MacBook Pro, for the work that actually takes time. On sustained, multithreaded tasks, the kind that use every core for minutes rather than one core for a fraction of a second, the M5 Max and the desktop Ryzen 9 and Core i9 chips remain dramatically ahead, simply because they have far more cores and can sustain higher power draw indefinitely.

✦ Free Newsletter ✦

Never miss a story

Tools, tutorials and AI deep-dives - straight to your inbox, every week.

Compiling a large codebase, exporting a 4K video timeline, running a local AI model, or rendering a 3D scene are all multi-core-bound tasks. On every one of those, the A20 Pro loses badly to the chip Apple put next to it in the same test, let alone to a desktop tower with a real cooling loop. Year over year, Apple's own numbers and Geekbench data show the A20 Pro improving roughly 21-23% in single-core and roughly 27-30% in multi-core over last year's A19 Pro, according to iGeeksBlog and wccftech, which is a genuinely solid generational gain. It just isn't the same claim as "beats desktop computers," and treating a single-core synthetic score as a stand-in for general performance is the exact kind of number that Geekbench 7 headlines are engineered to produce.

Who Should Actually Care About This Chip

This matters most to people upgrading a phone they use constantly for everyday bursts of activity: opening apps, scrolling, switching between messages and a browser tab, taking photos. Single-core speed genuinely does correlate with how snappy that kind of interaction feels, because most of what a phone does moment to moment is short, single-threaded work. If you're due for a new iPhone 18 Pro anyway, the A20 Pro is a legitimate reason to feel good about the purchase.

Close-up of a laptop keyboard and trackpad

It matters much less to anyone deciding between a phone and a laptop for real work, or comparing an iPhone to a gaming PC or a MacBook. If your workload is video editing, software compilation, or anything that saturates multiple cores for an extended stretch, the benchmark headline is not describing your use case, and you should look at multi-core and sustained-performance numbers instead, ideally from full independent reviews rather than early leaked single scores.

Choose to care about this benchmark if: you use a phone mainly for quick, everyday tasks and want to know the new model is a real generational step up.
Skip worrying about it if: you're comparing phones to laptops or desktops for sustained, multi-core work like editing, compiling, or rendering.
Wait for independent reviews if: you want a number you can actually trust, since early Geekbench leaks (this one included) vary by hundreds of points run to run.

What This Means Beyond One Phone Launch

The bigger story is competitive pressure on the rest of the chip industry. Apple's Arm-based, small-core, high-efficiency design philosophy keeps closing the gap with x86 desktop chips on single-thread speed, a trend that's been building since Apple moved Macs to its own silicon. Qualcomm and MediaTek both have their own 2-nanometer roadmaps in progress, and Windows-on-Arm laptops have been trying, with mixed results so far, to match what Apple's M-series Macs already do for battery life and quiet operation. A phone chip embarrassing a MacBook Pro chip on one metric is also, implicitly, a preview of what Apple might build into its next Mac silicon generation.

Rows of illuminated server racks in a data center

It's also a reminder that chip marketing has learned exactly which single number reliably generates headlines. GPU makers do something similar when they lean on frame-generation multipliers rather than raw rendered frames, as Nvidia has with its DLSS 5 marketing around the RTX 5090. A single benchmark score is real data, but it's data selected because it flatters a specific design choice, not because it's the most representative number available.

The Takeaway

Treat the A20 Pro's Geekbench 7 record as what it is: confirmation that Apple's newest phone chip has a genuinely fast pair of performance cores, built on a leading-edge process node, for the kind of short bursts that define everyday phone use. Don't treat it as evidence that your next iPhone can replace a laptop for real work, and don't take the exact point totals as gospel until full retail-unit reviews replace the early leaked runs. If you're buying the phone for how it feels day to day, the upgrade is real. If you're buying it to do heavier work, wait for the multi-core numbers and the actual reviews, not the single line that made the headlines.