When you think about chip performance, it is rare to see numbers that make you sit up. The first benchmark results for Apple's M5 chip do exactly that. The initial performance data from the 14-inch MacBook Pro has emerged, pointing to a leap in processing power that stretches what we expect from a laptop.
What stands out in these early results is how the M5 reads less like an incremental bump and more like a reboot of the rulebook. The benchmark shows a single-core score that is breaking new ground in the Geekbench 6 database, and the multi-core gains suggest a shift in what portable machines can realistically handle.
This is Apple's most aggressive swing at AI-focused computing so far, with architectural moves aimed at speeding up machine learning workflows and creative apps. The M5 is not just faster, it is tuned for AI-heavy tasks that pros now rely on, from real-time video enhancement to on-device model runs.
Record-breaking single-core performance sets new standards
About that single-core score, it is a head-turner. The M5 chip has achieved a single-core score of 4,263 points, which represents the highest score ever recorded in the Geekbench 6 database for any Mac or PC processor. Stack it against rivals and you get scores that rival Intel's Core Ultra 9 285K and surpass AMD's 9950X3D. That is desktop-class territory, in a laptop.
The architecture behind it is where the plot thickens. The chip features a 10-core CPU configuration with four performance cores and six efficiency cores, and Apple says its new performance core is the "world's fastest performance core". Not just marketing fluff, the numbers back it up, with results that outpace high-end desktops while staying cool enough for a thin-and-light.
Better yet, the consistency shows up across devices. The iPad Pro version of the M5 posts similar single-core figures, a reminder that Apple's silicon tuning delivers flagship speed whether you pick up a tablet or a laptop.
Multi-core gains deliver substantial improvements over M4
Multi-core tells a complementary story, with a wrinkle tied to thermals. With a score of 17,862 points, the M5 scales nicely and lands up to 20% faster performance compared to the previous M4 chip in the same 14-inch MacBook Pro. That puts the standard M5 in intriguing company, it is faster than the M3 Pro chip and nearly matches the performance of the M1 Ultra, a base chip delivering what used to be high-end numbers.
Thermals explain some cross-device gaps. The MacBook Pro's M5 running at 4.61 GHz shows a 9% performance advantage over the iPad Pro version running at 4.43 GHz. The difference tracks with cooling headroom, the MacBook Pro can sustain higher clocks, the iPad Pro keeps a thinner profile and prioritizes battery life.
For buyers, that split is practical. Long, heavy workloads like video renders or big builds will favor the MacBook Pro's steadier high frequencies, while the iPad Pro feels snappy for short bursts that match tablet habits.
AI and graphics performance receive major architectural upgrades
Here is where the M5 puts clear distance on prior chips. Apple reworked how AI and graphics talk to each other. The headline change, Neural Accelerators integrated in each GPU core, delivering over 4x the AI performance compared to M4. Not a mild tweak, a structural shift toward distributed AI units inside the graphics pipeline.
That distributed approach lifts graphics too. Performance sees up to 30% faster performance than M4 and 2.5x faster than M1, with support for more advanced rendering. A third-generation ray tracing engine provides up to 45% higher performance in ray-traced applications, which makes real-time ray tracing a realistic tool for creative work and some gaming on Mac.
Feeding all of that, bandwidth. The 153GB/s unified memory bandwidth represents nearly a 30% improvement over the M4's 120GB/s. More headroom means larger local AI models and higher resolution graphics without choking the pipeline.
What these benchmarks mean for real-world performance
The jump from pretty charts to actual time saved comes down to one theme, local AI horsepower. With the M5's AI muscle, many tasks that previously required cloud processing can now run locally on the device, which trims latency, sidesteps privacy worries, and keeps you productive offline.
For creatives, the upside shows up on the clock. Apple says the M5 delivers 7.7x faster AI video enhancement performance in Topaz Video compared to M1 systems. Paired with the bandwidth gains, you get better multitasking, like running Adobe Photoshop and Final Cut Pro while uploading large files to the cloud, with fewer pauses and less juggling.
The timing helps if you are eyeing an upgrade, pre-orders are available now and availability starts October 22. If you are on an Intel Mac or an early M-series machine, the M5 feels like a new class, not just a faster chip. Whether you are a developer poking at large language models, a video editor leaning on AI tools, or a designer experimenting with ray-traced scenes, these benchmark gains look like real minutes saved and new workflows unlocked. I would bet that once you try it, going back will feel slow.