Apple MacBook Pro M5 review: Incremental gains with a big AI leap
Apple’s newest MacBook may look unchanged, but under the hood, the M5 chip brings real gains for those building or running AI models. For everyone else, the improvements might feel more theoretical than transformative
Earlier this year, Apple marked the second consecutive year of launching a laptop with absolutely no identifiers of changes on the exterior. After a redesign in 2023, Apple’s MacBook Pro lineup is deemed to be fresh enough in design, and functional enough in practicality—prompting the company to make no changes to what is not broken. This year, the M5 chip’s advent made for even fewer changes—but for its target audience, this can be a crucial upgrade.
On paper, the M5 chip is clearly the better one. While its single-core performance is comparable with the M4 chip, the M5-powered MacBook Pro critically does well at improving its graphics processing unit (GPU) performance, as well as what the company calls ‘neural accelerators’ integrated into its GPU cores. Adding to all of this, the M5 processor now supports higher memory bandwidth of 153 gigabits per second.
All of this makes for a number of interesting real-world results. We’ll start with multitasking in everyday applications. If your work involves vast troves of low-latency data processing such as in banking, every split-second stutter would eventually begin showing up for you in large ways. The M4 chip, while being super-fast, showed this vulnerability—with two data terminals across service providers open in two separate browsers, there would be visible stutters even at high battery and low background processing levels.
These regular stutters have become nearly nonexistent now, with the M5 chip spearheading the processor lineup. Apple will of course have more variants of this chip in the form of M5 pro, max and ultra—all of which will be even better. But, as a basic standard line of performance, the M5 MacBook Pro ventures into a new territory that no previous base models did: specialized tasks.
For instance, with the M5 MacBook Pro at hand, I could easily download Gemma-3, a single-GPU-specialist small language model (SLM) from Google, and run it on the Ollama runtime application for AI models. The result was quite smooth and unfettered, and allowed me to create the basic code framework of a minimalistic notetaking application for MacBook platforms—all within one day’s work.
This, while possible to do on the M4, took much longer. Take the laptop off power, and the performance dip is very evident. This is where the M5’s actual capabilities show up—the neural accelerators in the GPU and the higher memory bandwidth mean that the laptop is able to recall data points in local storage in nearly 30% lesser time than before. Once data points are called by the AI model, the neural accelerators ensure that processing of AI algorithms is yet again faster than before.
But, a faster and better GPU may also mean a faster and better gaming laptop. While we couldn’t test it ourselves, we did see a brief, 15-minute demo of Cyberpunk 2077’s early build being played on an M5 MacBook Pro (with the base configuration of 16GB memory and 512GB SSD storage). Real-time ray tracing on the laptop looked sharp, and there were no evident frame drops in the very brief demo that I could see.
All of this beckons the question: who is this upgrade really for? And, do we really need MacBooks to be upgraded every year? After all, if someone spent a hefty penny two years ago on an M3 MacBook Pro, is it fair for them to have to upgrade within two years?
The answer is a bit more nuanced than a simple affirmative. You see, over the past two years, engineering tasks for artificial intelligence have truly ramped up. The MacBook Pro, on this note, was always built for ‘professional’ users who needed the extra firepower: instead of someone wanting a nicer-looking device, or with more cameras. For the professional clan, the upgrades between the M3 MacBook Pro and the M5 MacBook Pro would make a difference.
However, this said difference would only be evident for those working on cutting-edge large AI models, or highly complex applications being developed. Until then, the MacBook Pro’s M3 avatar would work well enough for those working on smaller projects—albeit at a pace that could at times be less than half of what the M5 is capable of working at. Would these users know that their laptop is slower than the latest in the market? Yes, but not empirically—most differences are truly evident comparatively.
The M5 MacBook Pro, however, is the best-suited laptop from Apple for the AI era. It will undoubtedly be better next year—and gaming isn’t quite its biggest use case yet. But, if you’re a developer or an enthusiast working closely on a new project, the M5 MacBook Pro is certainly the one to upgrade to. Users on the M1 MacBook Pro, released in November 2020, would particularly feel the jolt of speed—and would be the obvious candidates for an upgrade.
But on its own, the Apple MacBook Pro’s M5 edition—the latest for 2025—leaves no features amiss. The full-sized HDMI and the memory slots are retained, and three USB-C ports could be enough for most users choosing one for an external display, one for charging, and the other for a connector dock. The Retina display is as sharp as always, and the stereo speakers sound exactly the same as what Apple has offered in this new MacBook Pro design over the past two years. The keyboard, too, remains the same—and is comfortable enough to work on for both long hours, and at awkward angles.
At ₹1.7 lakh, is this a laptop you’d want to buy? In fact, we’d go as far as to say that if you’re a pro user dabbling with AI, this is the only laptop that you should buy today.
topics
