Apple M5 vs M4, M3, M2, M1 (+Pro/Max/Ultra): Full Comparison!

Apple has ingeniously designed its M-series CPUs to function like legos, allowing them to pair up and create more formidable chips. For instance, the M-Ultra chips are essentially two M-Max CPUs combined, effectively doubling their resources to challenge every Intel and AMD CPU. However, today’s focus isn’t on Team Blue and Team Red; instead, we’re comparing all four generations of M-series processors.

Additionally, we’ll provide projected scores for the Apple M5 Pro, M5 Max, and M5 Ultra, and of course for the missing M4 Ultra. Let’s go!

[01/2025] Updated: Added Apple M4 (8-core CPU), M4 (10-core CPU), M4 Pro (12-core CPU), M4 Pro (14-core CPU), M4 Max (14-core CPU), and M4 Max (16-core CPU)

[03/2025] Updated: Added Apple M3 Ultra

[11/2025] Updated: Added Apple M5 (9-core and 10-core)

[03/2026] Updated: Added A18 Pro, Apple M5 Pro (15-core and 18-core), and Apple M5 Max

Apple CPU lineup – The Full List

Each new generation of Apple chips brings significant upgrades, as Apple introduces more powerful and efficient architectures annually:

Apple M1 vs M2 vs M3 vs M4 vs M5

First, let’s examine the progression of the “base” variant of the M-series CPUs.

Below is a summary of the single-core performance for these base variants, showcasing performance scores and percentage improvements across successive generations.

Single-Core Performance

Single-core performance is pivotal for delivering optimal speed in applications that depend heavily on the processor’s core speed, such as gaming or software compilation. It also plays a significant role in determining the overall responsiveness and efficiency of operating systems and utility applications, ensuring smooth user experiences even when multi-core optimization is not available.

We’re adding the best-performing Intel chip in our Top Laptop CPU Ranking, the Intel Core Ultra 9 285HX (3124 points), for reference. It almost perfectly aligns with Apple M3. The Apple A18 Pro in MacBook Neo is just between M3 and M4.

The Apple M4 represents the most significant leap yet in single-core performance.

There are also two “trimmed” variants of M4 – the 8-core Apple M4 that can be found in the most budget iMac, and the 9-core Apple M4 that can be found in the most budget iPad Pro:

Multi-Core Performance

Multi-core performance is essential for handling complex and demanding tasks, such as video editing, 3D rendering, and scientific simulations.

The table below details the multi-core performance of Apple’s base M-series CPUs, highlighting score improvements and performance gains with each new generation. For context, we’ve added the top-performing Intel chip in our Top Laptop CPU Ranking, the Intel Core Ultra 9 285HX, which scores 22215 points.

The Apple A18 Pro in MacBook Neo is just above the Apple M1 chip.

Multi-core performance trends closely with single-core improvements, with the latest M5 marking the most substantial advancement. However, the base variants still trail behind the Intel flagship, Core Ultra 9 285HX, in complex tasks.

Again, here’s a M4 variants comparison:

Now, let’s compare this to the Pro, Max, and Ultra variants of the Apple M-series.

Apple M1 Pro vs M2 Pro vs M3 Pro vs M4 Pro vs M5 Pro

The 14-core M4 Pro has 10 performance cores that deliver a crazy performance jump of 45%! This is 3% higher than Intel’s mobile flagship Intel Core Ultra 9 285HX. So basically the Ultra 9 285HX is approximately as fast as the Apple M3 in single-core, and as the Apple M4 Pro in multi-core.

Apple M1 Max vs M2 Max vs M3 Max vs M4 Max vs M5 Max

M3 Max was the only Apple processor to outperform Intel’s then flagship i9-14900HX when comparing mobile processors. Now the blue flagship is Intel Core Ultra 9 285HX, and its multi-core result goes just between M3 Max and M4 Max.

M5 Max scores at 29 159 points, 31% above 285HX:

Apple M1 Ultra vs M2 Ultra vs M3 Ultra vs M4 Ultra and M5 Ultra (Projected)

Edit [03/2025]: Apple M3 Ultra was released much later than expected but it’s here, and it’s a huge update – 33% performance jump! This alters the projection for M4 Ultra from 33 773 pts to 37 867 pts, and the projection for M5 Ultra climbs to insane 42 864 pts!

Given that Ultra chips are typically used in desktop configurations, we’re swapping the mobile Intel Core Ultra 9 285HX with the desktop flagship Intel Core Ultra 9 285K which scores 22,000 points in the Multi-Core test and is 4% faster than the M2 Ultra.

As a result, M4 Ultra is expected to be 72% faster than Intel Core Ultra 9 285K and the M5 Ultra could be twice as fast as the Intel Core Ultra 9 285K. Of course, that’s just a projection, and Apple would probably need a redesign of the Mac Studio to be able to fit an M4 Ultra or M5 Ultra in it.

The Ultimate Comparison:

Now, let’s conduct a thorough comparison of all Apple M-series CPUs using actual data.

The latest Apple M5 processor is the fastest CPU in terms of single-core power:

When it comes to heavy work, the Apple M5 has multi-core peformance similar to the smaller variant of M3 Max (14-core) and it’s 14% below M4 Pro (12-core). However, the M4 Max variant boasts an extraordinary memory bandwidth (409.6 GB/s), which is 3-4x that of the M4 (120 GB/s) and 2x that of the M2 Pro (204.8 GB/s), proving beneficial for AI projects. The king for AI is the Apple M3 Ultra with 28,000+ multi-core score, and 819.3 GB/s memory bandwidth.

The M5 Pro variant matches the multi-core performance of M3 Ultra, and the M5 Max is even more powerful:

We’ve highlighted only the top-performing CPU variants above, where multiple versions of the same model exist. Here is the complete list of all M-series processors and their single and multi-core scores:

We hope this summary helps you decide which M-series CPU to choose.

You may also find our other analysis interesting:
The Gap is Getting Scary: Apple Leaves Intel in the Dust

And of course, don’t miss our comparison with the new exciting arrival, the Qualcomm Snapdragon X Elite:
Apple M4 vs Snapdragon X Elite – The Next-Gen CPU Battle

Author: Simeon NikolovEditor-in-Chief at LaptopMedia.com with 20+ years of experience in testing and analyzing laptops and mobile devices.

Subscribe

Notify of

new follow-up commentsnew replies to my comments

I allow to use my email address and send notification about new comments and replies (you can unsubscribe at any time).

{} [+]

Name*

Email*

Website

I agree to the Terms and Privacy Policy.

The comment form collects your name, email, and content to allow us keep track of the comments placed on the website. Please read and accept our website Terms and Privacy Policy to post a comment.

{} [+]

Name*

Email*

Website

I agree to the Terms and Privacy Policy.

The comment form collects your name, email, and content to allow us keep track of the comments placed on the website. Please read and accept our website Terms and Privacy Policy to post a comment.

22 Comments

Inline Feedbacks

View all comments

yesir

2 years ago

这个文章总结的太棒了，什么神仙编辑，把我想要的信息都罗列出来了

42

Reply

Author

Simeon Nikolov

2 years ago

Reply to  yesir

这条评论真是太温暖了，让我非常开心！谢谢你，Yesir！
—
That was the most heartwarming comment; it made my day! Thank you, Yesir!

Last edited 2 years ago by Simeon Nikolov

16

Reply

wu wei

1 year ago

Best Apple M chips analysis

7

Reply

Author

Simeon Nikolov

1 year ago

Reply to  wu wei

Thank you very much, wu wei!

1

Reply

abc

1 year ago

赞 赞 赞 加鸡腿

1

Reply

Author

Simeon Nikolov

1 year ago

Reply to  abc

🙂 👍

0

Reply

chee

1 year ago

incredible comparison design. any chance for gpu/npu?

6

Reply

Author

Simeon Nikolov

1 year ago

Reply to  chee

Good idea!

1

Reply

Wen

7 days ago

Reply to  Simeon Nikolov

Great article, would love to see update with GPU performance gain comparsion too

0

Reply

sikaco

1 year ago

Perfect analysis. There are all I want to get!

2

Reply

ntz

1 year ago

Hello, I don’t get the comparison with i9-14900K .. it’s just purposely written and depicted to make apple chips better … In reality i9 cpu that was used as a reference point is on the top of the list

0

Reply

Leo

1 year ago

Reply to  ntz

Check this out, bro, https://versus.com/en/intel-core-i9-14900hx-vs-intel-core-i9-14900k

0

Reply

Trashbean

2 months ago

Reply to  ntz

I had a look at your chart which does not include the M5 and the graphs above are accurate.
The i9-14900K only beats the lower spec Apple chips at multi core performance and Cinebench. It trails everywhere else.
You seem to be going off the Crossmark Overall score which is distorted by the large number of No Value results for the Apple chips because some of the tests do not run on the Macs.

0

Reply

Trashbean

2 months ago

Reply to  ntz

I had a look at your chart which does not include the M5 and the graphs above are accurate.
The i9-14900K only beats the lower spec Apple chips at multi core performance and Cinebench. It trails everywhere else.
You seem to be going off the Crossmark Overall score which is distorted by the large number of No Value results for the Apple chips because some of the tests do not run on the Macs.

0

Reply

WFR

6 months ago

Great effort for putting together such an extensive comparison. I only wish the Y-axis started at zero because at the moment, you’ve artificially amplified the true relative difference between all of them.

1

Reply

Author

Simeon Nikolov

6 months ago

Reply to  WFR

Thank you, WFR! While a zero-start axis is indeed the safest choice for showing absolute performance, my goal here was slightly different. I wanted to highlight the relative uplift between generations, using the M1 (2020) as a baseline, so readers can quickly judge “how big is the jump from what I have to what I’m considering” (with the added context of Intel CPUs as well). I initially tried charts starting at zero, but the bars became compressed and the differences between closer generations were much harder to read at a glance. For example, if you’re on M4 and mostly care… Read more »

Last edited 6 months ago by Simeon Nikolov

1

Reply

Trashbean

2 months ago

Reply to  Simeon Nikolov

You do not show the relative uplift if you chop off the lower part of the graph.

For relative values you HAVE to show the whole enchalada, not the clipped M4 minus M1 which make the M1 zero value.

The % uplift you show in the figures actually works off the WHOLE value not a clipped value.

Last edited 2 months ago by Trashbean

0

Reply

Alfred

5 months ago

It would be great to have all M series Apple silicon GPU benchmarks.

0

Reply

tatachi

4 months ago

每次都看这个总结，各种对比数据太清晰了，期待m5pro和max的加入，感谢你的付出

1

Reply

Author

Simeon Nikolov

4 months ago

Reply to  tatachi

Thank you! The new models are coming to our lab, stay tuned for full analysis and reviews!
—-
谢谢！新模型即将送达我们的实验室，敬请期待完整的分析和评测！

0

Reply

Trashbean

2 months ago

Amazing! I was just about to do a spreadsheet on exactly this and found this site immediately. Thanks for the great work. 😃👏 All that’s missing is a “Bangs for Bucks chart” with the performance rated against price for each model. Just one suggestion. The Charts are too big for the information. If they were smaller you could take in all the information withouy scrolling. You should always start charts them from zero clipping them does not show the correct relativity. For the pricing I’d tilt the charts to axonometric view with $ as the height off the chart. We… Read more »

0

Reply

lastonesky

1 month ago

It would be even better if there were a graphics performance rating.

0

Reply