Wednesday, October 15th 2025
Apple Unleashes M5, the Next Big Leap in AI Performance for Apple Silicon
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 153 GB/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.
"M5 ushers in the next big leap in AI performance for Apple silicon," said Johny Srouji, Apple's senior vice president of Hardware Technologies. "With the introduction of Neural Accelerators in the GPU, M5 delivers a huge boost to AI workloads. Combined with a big increase in graphics performance, the world's fastest CPU core, a faster Neural Engine, and even higher unified memory bandwidth, M5 brings far more performance and capabilities to MacBook Pro, iPad Pro, and Apple Vision Pro."
A Next-Generation GPU Architecture Optimized for AI and Graphics
With the next-generation GPU architecture in M5, every compute block of the chip is optimized for AI. The 10-core GPU features a dedicated Neural Accelerator in each core, delivering over 4x peak GPU compute compared to M4, and over 6x peak GPU compute for AI performance compared to M1. And now with M5, the new 14-inch MacBook Pro and iPad Pro benefit from dramatically accelerated processing for AI-driven workflows, such as running diffusion models in apps like Draw Things, or running large language models locally using platforms like webAI.
The next-generation GPU and enhanced shader cores in M5 also deliver increased graphics performance, achieving up to 30 percent faster performance compared to M4 and up to 2.5x faster performance than M1. M5 also includes Apple's third-generation ray tracing engine, providing up to a 45 percent graphics uplift in apps using ray tracing. Combined with rearchitected second-generation dynamic caching, the GPU provides smoother gameplay, more realistic visuals in 3D applications, and faster rendering times for complex graphics projects and other visually intensive applications. With M5, Apple Vision Pro renders 10 percent more pixels with the micro-OLED displays, and refresh rates increase up to 120 Hz, resulting in crisper details, more fluid display performance, and reduced motion blur.
The GPU architecture is engineered for seamless integration with Apple's software frameworks. Applications using built-in Apple frameworks and APIs—like Core ML, Metal Performance Shaders, and Metal 4—can automatically see immediate increases in performance. Developers can also build solutions for their apps by directly programming the Neural Accelerators using Tensor APIs in Metal 4.
A Faster Neural Engine to Power Intelligent Features
The faster 16-core Neural Engine delivers powerful AI performance with incredible energy efficiency, complementing the Neural Accelerators in the CPU and GPU to make M5 fully optimized for AI workloads. For example, AI-powered features on Apple Vision Pro—like the ability to transform 2D photos into spatial scenes in the Photos app, or generating a Persona—operate with greater speed and efficiency.
The Neural Engine in M5 also enhances performance for Apple Intelligence. On-device AI tools like Image Playground get faster, and the overall performance of Apple Intelligence models are enhanced by the faster Neural Engine and unified memory in M5. Also, developers using Apple's Foundation Models framework will get faster performance.
Enhanced Memory to Do Even More with AI
M5 offers unified memory bandwidth of 153 GB/s, providing a nearly 30 percent increase over M4 and more than 2x over M1. The unified memory architecture enables the entire chip to access a large single pool of memory, which allows MacBook Pro, iPad Pro, and Apple Vision Pro to run larger AI models completely on device. It fuels the faster CPU, GPU, and Neural Engine as well, offering higher multithreaded performance in apps, faster graphics performance in creative apps and games, and faster AI performance running models on the Neural Accelerators in the GPU or the Neural Engine. And with 32 GB of memory capacity, M5 also helps users to seamlessly run demanding creative suites like Adobe Photoshop and Final Cut Pro simultaneously, while uploading large files to the cloud in the background.
Apple Silicon and the Environment
Apple 2030 is the company's ambitious plan to be carbon neutral across its entire footprint by the end of this decade by reducing product emissions from their three biggest sources: materials, electricity, and transportation. The power-efficient performance of M5 helps the new 14-inch MacBook Pro, iPad Pro, and Apple Vision Pro meet Apple's high standards for energy efficiency, and reduces the total amount of energy consumed over the product's lifetime.
Apple revolutionized personal technology with the introduction of the Macintosh in 1984. Today, Apple leads the world in innovation with iPhone, iPad, Mac, AirPods, Apple Watch, and Apple Vision Pro. Apple's six software platforms—iOS, iPadOS, macOS, watchOS, visionOS, and tvOS—provide seamless experiences across all Apple devices and empower people with breakthrough services including the App Store, Apple Music, Apple Pay, iCloud, and Apple TV. Apple's more than 150,000 employees are dedicated to making the best products on earth and to leaving the world better than we found it.
Source:
Apple
"M5 ushers in the next big leap in AI performance for Apple silicon," said Johny Srouji, Apple's senior vice president of Hardware Technologies. "With the introduction of Neural Accelerators in the GPU, M5 delivers a huge boost to AI workloads. Combined with a big increase in graphics performance, the world's fastest CPU core, a faster Neural Engine, and even higher unified memory bandwidth, M5 brings far more performance and capabilities to MacBook Pro, iPad Pro, and Apple Vision Pro."
With the next-generation GPU architecture in M5, every compute block of the chip is optimized for AI. The 10-core GPU features a dedicated Neural Accelerator in each core, delivering over 4x peak GPU compute compared to M4, and over 6x peak GPU compute for AI performance compared to M1. And now with M5, the new 14-inch MacBook Pro and iPad Pro benefit from dramatically accelerated processing for AI-driven workflows, such as running diffusion models in apps like Draw Things, or running large language models locally using platforms like webAI.
The next-generation GPU and enhanced shader cores in M5 also deliver increased graphics performance, achieving up to 30 percent faster performance compared to M4 and up to 2.5x faster performance than M1. M5 also includes Apple's third-generation ray tracing engine, providing up to a 45 percent graphics uplift in apps using ray tracing. Combined with rearchitected second-generation dynamic caching, the GPU provides smoother gameplay, more realistic visuals in 3D applications, and faster rendering times for complex graphics projects and other visually intensive applications. With M5, Apple Vision Pro renders 10 percent more pixels with the micro-OLED displays, and refresh rates increase up to 120 Hz, resulting in crisper details, more fluid display performance, and reduced motion blur.
The GPU architecture is engineered for seamless integration with Apple's software frameworks. Applications using built-in Apple frameworks and APIs—like Core ML, Metal Performance Shaders, and Metal 4—can automatically see immediate increases in performance. Developers can also build solutions for their apps by directly programming the Neural Accelerators using Tensor APIs in Metal 4.
A Faster Neural Engine to Power Intelligent Features
The faster 16-core Neural Engine delivers powerful AI performance with incredible energy efficiency, complementing the Neural Accelerators in the CPU and GPU to make M5 fully optimized for AI workloads. For example, AI-powered features on Apple Vision Pro—like the ability to transform 2D photos into spatial scenes in the Photos app, or generating a Persona—operate with greater speed and efficiency.
The Neural Engine in M5 also enhances performance for Apple Intelligence. On-device AI tools like Image Playground get faster, and the overall performance of Apple Intelligence models are enhanced by the faster Neural Engine and unified memory in M5. Also, developers using Apple's Foundation Models framework will get faster performance.
Enhanced Memory to Do Even More with AI
M5 offers unified memory bandwidth of 153 GB/s, providing a nearly 30 percent increase over M4 and more than 2x over M1. The unified memory architecture enables the entire chip to access a large single pool of memory, which allows MacBook Pro, iPad Pro, and Apple Vision Pro to run larger AI models completely on device. It fuels the faster CPU, GPU, and Neural Engine as well, offering higher multithreaded performance in apps, faster graphics performance in creative apps and games, and faster AI performance running models on the Neural Accelerators in the GPU or the Neural Engine. And with 32 GB of memory capacity, M5 also helps users to seamlessly run demanding creative suites like Adobe Photoshop and Final Cut Pro simultaneously, while uploading large files to the cloud in the background.
Apple Silicon and the Environment
Apple 2030 is the company's ambitious plan to be carbon neutral across its entire footprint by the end of this decade by reducing product emissions from their three biggest sources: materials, electricity, and transportation. The power-efficient performance of M5 helps the new 14-inch MacBook Pro, iPad Pro, and Apple Vision Pro meet Apple's high standards for energy efficiency, and reduces the total amount of energy consumed over the product's lifetime.
Apple revolutionized personal technology with the introduction of the Macintosh in 1984. Today, Apple leads the world in innovation with iPhone, iPad, Mac, AirPods, Apple Watch, and Apple Vision Pro. Apple's six software platforms—iOS, iPadOS, macOS, watchOS, visionOS, and tvOS—provide seamless experiences across all Apple devices and empower people with breakthrough services including the App Store, Apple Music, Apple Pay, iCloud, and Apple TV. Apple's more than 150,000 employees are dedicated to making the best products on earth and to leaving the world better than we found it.
28 Comments on Apple Unleashes M5, the Next Big Leap in AI Performance for Apple Silicon
Probably a few of each, but you never know when it comes to the fruity boys :D
Also, 15% uplift in MT performance is just...ok. Nothing revolutionary.
These days MacOS lags behind Linux while the M series processors are above average. Now people buy the hardware in spite of the software.
Their transition from TSMC's 2nd generation to 3rd generation 3nm process node basically accounts for the ~10%+ increase in single-core performance improvement in one synthetic benchmark (Geekbench).
There is a slightly larger increase in multi-core performance (like ~15%+) in the same benchmark suite. Apple introduced new architecture for their efficiency cores which do more work at the same power level. I also read something about improved thread scheduling.
Their GPU cores now have "neural accelerators" which is the source of much of their more substantial AI performance improvements on the GPU cores.
And I believe Apple made some improvements to their Neural Engine (ML) cores. The latter two combined make up the big leap in AI performance.
I am not sure what else they have done on the software side.
I'm pretty sure that the M5 doesn't do anything that the M1, M2, M3, and M4 SoCs can't do. The M5 is just faster/more power efficient. Apple seemingly has not tied any (major) feature to a specific generation of M-series silicon.
There might be some better support for hardware media encoding/decoding (higher resolution, higher framerates, greater bit depth) in the newer chips. This is not something that I would notice since I'm not a professional videographer but serious content creators using Premiere, Final Cut, Resolve, etc. to edit ProRes RAW probably track these sort of improvements.
It's also possible that some of the newer M-series SoCs also have hardware raytracing capabilities that the M1, M2 don't have but again I'm not sure of this. Absence of RT cores isn't a dealbreaker since Macs aren't purchased with gaming as the primary usage case.
It's hard to say exactly what Apple's semiconductor M.O. is but to me (not a chip guy) it appears that Apple isn't trying to build an entirely new chip at once but periodically introduces new technologies in the various core types (CPU performance, CPU efficiency, GPU, ML). That probably is easier on the software side as well.
It's crystal clear now that Apple takes a very holistic view about chip design and efficiency with the introduction of the C1/C1X (cellular data) and N1 (wifi & Bluetooth) silicon. It's no longer just about the main SoC (A- and M-series CPU/GPU/ML) but extends to other custom silicon. There are rumors about chiplets in Apple Silicon which may eventually mean putting this communications silicon in the package.
All devices using M-series SoCs have 8GB or more of RAM so they all support Apple Intelligence (in its current form) if the user wants to enable that. Apple Intelligence is disabled by default and the user is prompted by Apple whether or not they want to enable it during the initial device setup sequence. It can be enabled later with a simple toggle in the System Settings control panel.
Some of the geekier chip-focused sites will inevitably have a deeper dive into the M5's capabilities in the weeks after these systems ship (on the 21st of this month). Apple did not announce any desktop products with the M5 chip today (Mac mini, iMac) so we won't know if there are some thermal constraints once we see third-party MacBook Pro benchmark results. But presumably there may be some performance differences between the iPad Pro M5 and the MacBook Pro M5 from their differing thermal capabilities.
In a month or so you should have more answers to your various questions about the M5 silicon either here or elsewhere on the Internet.
Maybe semi accurate or chips and cheese will make them at some point
I keep trying Linux, both Gnome and KDE, normally under Fedora and I always and I mean always bump into these tyiny but annoying issues that they should not exist and were already fixed a long time ago by MacOS and Windows.
One of the many, bluetooth devices, especially audio like headphones.
Or Wayland still having issues with windows decorations, programs being properly displayed and other things.
I have been on that rollercoaster for over 20 years and even though Linux has advanced in leap and bounds, it is still behind in many things were it shouldn't.
Then again, per the myriad of negative posts, looks like Tahoe is a mess that shouldn't had been released so that's that.
I'm also fairly certain that Apple's graphics engineers (hardware AND software) have a pretty good grasp about the various usage cases for RT cores versus regular GPU raster cores versus CPU cores versus ML cores. If they really wanted to go after the offline 3D market, most likely they would have focused on developing the relevant silicon years ago. It's not like offline 3D rendering is a new use that popped up in the past several years and Apple already has a history in this field (c.f., RenderMan).
Apple's commitment to gaming on the Mac is pretty obvious. They deprecated 32-bit executables as well as OpenGL. They have switched system architectures TWICE this century (PowerPC > Intel > Apple Silicon). They have also deprecrated their two emulation utilities (Rosetta and Rosetta 2) so it's really hard to play older Mac games on a 2025 Mac running Tahoe/Sequoia/Sonoma (the three versions of macOS that are receive support).
Apple is also familiar with the dominant gaming APIs and understands how its four generations of the Metal graphics API fit into the gaming world. Generally speaking they care about games on iPhones and iPads, not so much for Macs. They are happy when something like Cyberpunk 2077, Resident Evil, Baldur's Gate III release for Mac but they aren't going to underwrite development in any sort of significant way.
They essentially conceded the 3D CAD/CAM market to Windows/x64 in the Nineties as did many others at the time (SGI, Sun, etc.). Apple probably has a good idea of how much the offline 3D hardware market is worth today and what sort of revenue they could realistically hope to carve out if they attempted to jump back in. It's not like they would wrest every single workstation away from PC/Windows.
If they really wanted to pump Apple Silicon full of RT cores, they would. Remember that in addition to the base M_ SoC, they also offer Pro, Max, and occasionally Ultra variants. All of those can have different core counts so they could stick more RT cores in the Pro/Max/Ultra designs too. If you are waiting for Apple to pump their M-series SoCs full of RT cores, you are probably going to be waiting a very looooong time unless Apple finds a new workload (AI related, probably) that would exploit RT silicon.
Nothing I've said here is revelatory. Apple's silicon today is defined by their business focus from the past 30 years, a business focus that has shifted and adapted over time because NOTHING stands still. It's not like they come up with one design scrawled on a whiteboard, tape it out, and ship it.
They've been specifically promoting games for MacOS for several years now, which is a major change from their posture in the 2010s when they could care less whether any game ran on Macs, since they only had Intel iGPUs in most of them.
Remember that they announce A-series silicon first. It's helpful for the Macs to have the same graphics technology for iOS game developers to write their titles. It also helps in bringing back some iPad software to Mac now that they all run on the same basic silicon.
But Apple's support of Metal, etc. on the Mac are primarily so there are tons of games on the iOS App Store.
Money talks and for Apple and gaming, it is overwhelmingly on iOS/iPadOS. No one is going to buy a Mac Studio to play CP2077. If it runs well, great, no one at Apple is going to complain about that.
I have a rather large Steam library (currently 360 games) and I put any Mac-supported games in a special Apple folder: there are 26 games in the Mac folder so less than 8% of these Steam titles will run on Mac. I expect this percentage to decrease over the years. And no, I am not crazy enough to buy Mac games on the Mac App Store, better to buy them on Steam and have multi-platform.
Apple recognizes that gaming is a powerful branding tool, even if the gaming they get from desktop gaming will be negligible compared to the umpteenth Candy Crush clone. If you can show off your hardware playing lots of demanding games, that's great advertising, even if most people who buy a Mac will never play Cyberpunk. It's similar to how Nvidia always highlights the performance of the 4090 and 5090, even though 99.5% of gamers will never have an xx90 GPU. Showing your best-case scenario is the way to advertise.
But that doesn't make Mac a viable and growing gaming platform. Hell, Apple doesn't even support Vulkan directly. If they were serious about Mac gaming, they'd definitely pursue that. Even wimpy Switch supports Vulkan.
Apple offers a wide array of tools at game developers' disposal. And they are using these tools for iOS/iPadOS gaming but balking on doing the work for macOS. The Mac market is too small and probably always will be. Moreover with >85% of Macs sold being notebook models, there are even fewer high-specced desktop Mac boxes. Sure, I've downloaded a couple of those 26 Mac games from Steam and run them on my Mac mini M2 Pro to check out performance. They run okay but not as good as my Windows gaming build (Ryzen 5800X3D CPU and RTX 4080 Super GPU).
Remember that most Mac households also have a Windows PC under the same roof. And some also have videogame consoles (PlayStation, Xbox, Switch).
Apple is all about being the performance-per-watt king, they aren't chasing the FPS benchmark crown.
And when you are looking at Apple, remember that the iPhone is their cash cow.
We've covered this many, Many, MANY times before: ten years ago the smartphone dethroned the personal computer as the primary driver of all consumer electronics innovation. Not just for Apple. FOR EVERYONE. The personal computer -- while a powerful and important tool -- is no longer the place to look at to see where consumer technology is going.
Hell, all of the AI chatbot stuff is smartphone focused and most of it debuts on iPhone before Android. Even ChatGPT's new generative video tool Sora still isn't on Android while iPhone users on the free service just got a bump in video length. Yes, there will always be some whiz-bang stuff on PCs that smartphones can't do. But PCs simply aren't the platform for introducing innovation anyhow. Even Macs don't have the Visual Intelligence feature yet: it's still only on iPhones despite the fact that most Macs (their notebooks) have a built-in camera.
Many people here cling tightly to their widescreen desktop monitor, ATX mid-tower box at their feet, and mechanical keyboard. Great, you guys are defending your benchmark crowns. But the rest of the world walked out the front door with their smartphone in hand. That's where anyone involved in consumer technology wants to be: in that person's smartphone.
I've owned a notebook computer since 2002, a smartphone since 2011, a tablet from around the same time and it's crystal clear when I travel that I'm spending far less time on the computer these days, much more time on the tablet and phone. The world has changed. I don't expect to live a computer-free life for the foreseeable future but I do expect my overall computer screen time to continue dropping.
Computers aren't useless but with each passing month they are increasingly becoming devices for niche workloads.
Around that time a bunch of blogspot, podcast about switching to windows became common. Those who are still attached to MacOS are still on the lookout and you'll see article wondering if the new mac pro, mac studio is worthy of a switch back.
Offline GPU rendering gained traction right when Apple started beefing with Nvidia and eventually stopped using their hardware. Otoy octane was released in 2011, Redshift 3D was released in 2014, Vray in 2014 as well. Arnold and Renderman GPU support are very young, and still not fully considered production ready yet compared to the CPU version. Even for nvidia hardware accelerated support for RT in offline rendering is very recent, that started around 2019. Jobs started the groundwork when he pushed for OpenCL, but Apple dropped the ball. Blender devs mentionned how bad open CL performance was on the mac even with equivalent windows hardware. MacOS would stay without proper GPGPU API support until metal.
Some studio stayed loyal for a while, like lunaranimation, an highend 3D animation studio working for big clients. From what I could see they stayed on the mac until ~ 2024. RTX accelerated workflow became too good I guess.
www.lunaranimation.com/lunar-blog/2021/03/02/mac-pro-a-year-in-the-studio
They had a commercial operation with Otoy to give a free year of octane X to new mac owners with apple silicon (Otoy who's still ignoring AMD hardware) and they got maxon to support Metal a year before AMD could get HIP support. Vray also added support for RT rendering on the mac, while still wanting nothing to do with AMD. Their marketing material and first party benchmark also support those effort made with those 3d software developers.
The improvement in RT offline rendering is healty, and far bigger than the improvement in gaming. I don't see them going back in heavy 3D production, but they are sucking less and less for a freelancer with humbler rendering needs.
While they have not explicitly stated so, I think Apple primarily sees Macs as a tool that services other devices/platforms: iPhone, iPad, Apple TV, Apple Watch, Apple Vision Pro, etc. particularly in the content creation realm (films, television, games, music, etc.). I don't think they really see the Mac as the primary device that people will be consuming this content on here in 2025.
As Steve Jobs described the iPhone in 2007 at its unveiling: "It's the computer for the rest of us."
Macs are for content creation. iPhone, iPad, Apple TV, AirPods, etc. are for content consumption. Twenty years ago people used computers for Internet content consumption because we didn't have anything else. The world has changed since then and computers are maybe the 4th, 5th, 6th, 7th choice for content consumption.
Apple realistically knows that for AAA PC videogame development, a Mac isn't going to be the start-to-finish solution. At best, it will be part of the workflow of hundreds even thousands of computers, if the game isn't an iOS-focused title. Naturally the 3D graphics world continues to evolve and all players (hardware and software) need to adapt to the ever changing landscape.
Clearly Apple is okay with the Mac not being a destination for consumer content consumption. And apart from gaming the average Windows PC is also a pretty bad consumer content consumption device as well. Microsoft basically has written themselves out of the consumer story (with the exception of Xbox) because they have fumbled every single consumer initiative they have undertaken.
As long as Apple sells oodles of iPhones the Mac will definitely have an important place in Apple's product lineup as an iOS developer's tool.