Apple M5, M5 Pro and M5 Max: Which MacBook Chip Do You Need?

The three chips at a glance

Apple builds the M5 family in three steps. All three share the same basic design: a CPU, a GPU, a 16-core Neural Engine and one pool of unified memory on the same package. What changes from step to step is how many CPU and GPU cores you get, how much memory the chip can address, how fast that memory is (bandwidth), and how many video engines and external displays it supports.

The table below uses only the figures Apple prints on its MacBook Air and MacBook Pro tech specs pages. Apple names the CPU cores in a new way for this generation: the M5 pairs super cores with efficiency cores, while the M5 Pro and M5 Max pair super cores with performance cores. That is a meaningful difference in itself, since the higher chips trade the low-power efficiency cores for more cores built for sustained work.

Which MacBook carries which chip

The chip also decides which MacBook you can buy, because Apple does not offer every chip in every model. According to Apple's tech specs pages, the lineup splits like this:

• MacBook Air 13-inch: M5 with an 8-core GPU and 16GB of unified memory, configurable to 24GB or 32GB.
• MacBook Air 15-inch: M5 with a 10-core GPU and the same 16GB, 24GB or 32GB memory choices.
• MacBook Pro 14-inch: M5 (10-core GPU, 16GB to 32GB), M5 Pro (15-core or 18-core CPU) or M5 Max (32-core or 40-core GPU).
• MacBook Pro 16-inch: M5 Pro with the 18-core CPU and 20-core GPU, or M5 Max. There is no plain M5 option in the 16-inch model.

CPU cores: what the Pro and Max change

Every M5 chip starts with the same kind of super core, so tasks that lean on one or two threads, such as opening apps, browsing, office documents and most everyday software, feel similar across the family. The differences show up when many cores are busy for minutes at a time: compiling large code bases, exporting long video timelines, batch processing photos or running several virtual machines.

The M5 has 10 cores, six of them efficiency cores tuned for low power. The M5 Pro moves to 15 or 18 cores, and the extra cores are performance cores rather than efficiency cores. The M5 Max uses the same 18-core CPU as the top M5 Pro. That means the step from the 18-core M5 Pro to the M5 Max buys you GPU, memory and media engines, not more CPU cores. If your heavy work is mostly CPU-bound, such as software builds or data processing, the 18-core M5 Pro is the chip to look at.

GPU cores and memory bandwidth: video, 3D and local AI

The GPU is where the three chips differ most. The M5 tops out at 10 GPU cores, the M5 Pro at 20 and the M5 Max at 40. Memory bandwidth climbs in step: 153GB/s on the M5, 307GB/s on the M5 Pro, and 460GB/s or 614GB/s on the M5 Max depending on the GPU. On Apple silicon the GPU reads directly from unified memory, so bandwidth is as important as the core count for GPU-heavy work.

More GPU cores and bandwidth help with work that keeps the GPU busy: color grading, effects and noise reduction in video editors, 3D rendering and viewport work, and running large language models locally. For local AI in particular, generating each token means reading through the model weights, so higher bandwidth generally means faster output, and more memory means a larger model fits at all. Our laptop specs for local AI article explains how to size memory for a given model.

The M5 Max also doubles the media hardware: Apple lists two video encode engines and two ProRes encode and decode engines, against one of each on the M5 Pro. That helps with exports and with editing several streams of ProRes footage at once. The video editing laptops guide compares machines for that kind of work.

What the bigger chips do not change

Several things stay the same across the family, so do not pay for a higher tier expecting them to improve. All three chips have a 16-core Neural Engine, so on-device features in macOS that run on the Neural Engine do not get a bigger engine on the Pro or Max. Everyday responsiveness in a browser, mail or office apps is set by the single super core, which every chip has.

The hardware video formats are the same too: every chip lists hardware acceleration for H.264, HEVC, ProRes and ProRes RAW, plus AV1 decode. The Max adds more engines, not new formats. And a bigger chip does not make a small job finish faster if the job never keeps the extra cores busy. For light editing, school work and web development, the M5 already does the job, and the extra GPU cores of an M5 Max would mostly sit idle.

Finally, the chip does not change the operating system. If you are still weighing macOS against Windows, read MacBook or Windows laptop before choosing a tier.

Configure-to-order memory: choose it at purchase

Unified memory is part of the chip package, so it cannot be upgraded later. Apple sells a standard amount and lets you configure more at order time, and the memory you can choose depends on the exact chip and GPU you pick.

Apple's 16-inch MacBook Pro tech specs spell out the rules: the M5 Pro comes with 24GB and can be configured to 48GB, or to 64GB with the 20-core GPU. The M5 Max with the 32-core GPU comes with 36GB. The 48GB, 64GB and 128GB options on the M5 Max require the 40-core GPU, which is also the configuration with 614GB/s of bandwidth. In practice, 128GB is only available on the top M5 Max.

On the M5 side, the MacBook Air starts at 16GB and is configurable to 24GB or 32GB, and the 14-inch MacBook Pro with M5 offers 16GB, 24GB or 32GB. Because none of this can change after purchase, lean one tier higher than you would on a laptop with memory slots. Apple lists 8GB of memory as the minimum for Final Cut Pro and recommends 16GB, while Compressor needs 32GB (64GB recommended) to process Apple Immersive Video. Our RAM guide covers memory needs for other kinds of work.

How to choose between M5, M5 Pro and M5 Max

Start from the work, not the chip name. If you mostly write, browse, study, code web apps and edit photos, the M5 is enough, and it is the only chip in the MacBook Air. The Apple M5 laptops guide compares those machines. Choose the memory first (24GB or 32GB if you keep the laptop for years), then the model.

If you edit video regularly, build large projects or want three external displays, the M5 Pro is the balanced choice. It keeps the full 18-core CPU of the top tier, doubles the M5's bandwidth and goes to 64GB of memory with the 20-core GPU.

If the GPU is your bottleneck, such as heavy video grading, 3D rendering, multiple ProRes streams or local AI models that need more than 64GB, go to the M5 Max, and to the 40-core GPU if you need more than 36GB. The Apple M5 Max laptops guide lists current configurations. To see the whole lineup side by side, start with the Apple laptops or macOS laptops guides.

What to pick for your use

Questions

Is the M5 Pro worth it over the M5?

Yes if you keep many cores or the GPU busy for long stretches, such as video exports or large code builds. Apple lists up to 18 CPU cores, up to 20 GPU cores, 307GB/s of bandwidth and up to 64GB of memory for the M5 Pro, against 10 CPU cores, up to 10 GPU cores, 153GB/s and up to 32GB for the M5. For everyday work the M5 is enough.

Does the M5 Max have a faster CPU than the M5 Pro?

Not in core count. Apple lists the same 18-core CPU (6 super cores and 12 performance cores) for the top M5 Pro and for the M5 Max. The Max adds GPU cores, memory bandwidth, larger memory options and a second video encode and ProRes engine.

Which chip do I need for 128GB of unified memory?

The M5 Max with the 40-core GPU. According to Apple, the M5 Max with the 32-core GPU comes with 36GB, while the 48GB, 64GB and 128GB options require the 40-core GPU. The M5 Pro tops out at 64GB.

Can I upgrade the memory of an M5 MacBook later?

No. Unified memory is part of the chip package and is chosen when you order. Configure the amount you will need for the life of the laptop.

Can I get an M5 Pro or M5 Max in a MacBook Air?

No. Apple offers the MacBook Air only with the M5. The M5 Pro and M5 Max are available in the 14-inch and 16-inch MacBook Pro, and the 16-inch model has no plain M5 option.

Does the M5 Max make local AI faster?

For large models, usually yes, because it has the most memory and the most bandwidth in the family, and token generation depends heavily on reading weights from memory. For small models that fit easily in 32GB, the gap narrows, and the choice of model size matters more than the chip.