Some of the links in this post are affiliate links, which means I may earn a small commission if you choose to make a purchase at no extra cost to you. As an Amazon Associate, I earn from qualifying purchases. Read my full Ethics & Affiliate Relationship Disclaimer.
The M5 MacBook Pro still starts with 16GB of RAM, and I keep getting the same question: is it finally time to start the Pro at 24GB instead? So I put the two side by side. One is the M5 MacBook Pro with 16GB of RAM and a 512GB SSD. The other is the 24GB model, also with a 512GB SSD. I wanted to see the real difference between them and figure out whether spending a little more on memory does more for you than you would expect.
The 16GB base model holds up better than you might think, but the roughly $200 jump to 24GB buys a bigger buffer that pays off in AI work, heavy multitasking, and a couple more years of useful life.
- About $200 less
- Handles everyday multitasking well
- Most benchmarks were close, and everyday browsing barely feels different
- Hits yellow memory pressure much later, 35 tabs vs 20
- Dramatically faster for local AI in LM Studio
- A bigger buffer for a couple more years of use, since RAM can’t be upgraded later
Who this is for
This is for anyone standing in front of the M5 MacBook Pro configuration page trying to decide whether to click the 24GB option. If you mostly browse, write, and run a handful of apps, the 16GB base is going to feel fine. If you push into local AI models, video editing, 3D work, or heavy photo exports, this is where the extra memory starts to earn its keep. I tested both machines with the same tasks so you can see exactly where the line falls.
| 16GB model | 24GB model | |
|---|---|---|
| Chip | M5, 10-core CPU, 10-core GPU | M5, 10-core CPU, 10-core GPU |
| RAM | 16GB | 24GB |
| Memory bandwidth | 153GB/s | 153GB/s |
| SSD (as tested) | 512GB | 512GB |
| Upgrade cost | Base | About $200 more |
On paper, RAM is the only difference between these two. Both of my test units had a 512GB SSD. Apple has since stopped selling the 512GB model, and the 14-inch M5 MacBook Pro is now 1TB or more.
What unified memory does
Before going further, it helps to know what the RAM is doing. Your CPU is incredibly fast and can write tons of data every second. The SSD is slow by comparison. So the RAM acts as a buffer where the CPU can quickly write things that need to be remembered temporarily. When something needs long-term storage, the computer moves it to the SSD.
The way I think about it, having more RAM is like having a bigger workbench. The more space you have on the bench, the more tasks you can get done without bumping into tools and materials that are in your way. More RAM keeps more of your work readily available for the CPU, which improves how quickly you can switch between apps, or even between tasks inside the same app. When you start to run low, the computer falls back on the SSD for those tasks, and that is when you feel the slowdowns.
Memory pressure
For plain day-to-day use, running multiple browser tabs and doing normal multitasking, I did not really notice the 24GB feeling all that much faster than the 16GB. The difference shows up when you push more intensive apps or drive the memory pressure higher.
So I opened Safari and Activity Monitor and started stacking tabs to see how long it would take to reach yellow memory pressure. On the 16GB machine, it only took 20 tabs to get into the yellow. On the 24GB machine, it took 35 tabs. Then I pushed both up to 50 Safari tabs. At that point the 16GB one was pretty mad at me. It was really not responding well at all. The 24GB machine had a lot more room before it started to struggle.
The local AI test
Then I opened LM Studio, which is a pretty intensive app that lets you run AI models locally on the computer. I loaded GPT-OSS on both machines. Before the 16GB one could even load the model, it was already loaded on the 24GB machine. I typed in a question and got a response with about five seconds to the first token on the 24GB. The 16GB one just kept spinning for minutes. I finally got tired of waiting and canceled the test.
That test told me something clear. If you really want to push these machines, running video editing or 3D rendering software and switching back and forth between apps, that is when you are going to appreciate the extra memory. You can go all the way up to 32GB on the M5 if you want, but even stepping up to 24GB made a huge difference here.
Time to first token and other tests
Back in LM Studio with GPT-OSS, the big difference was the time to first token, which is how long the computer takes to process the request you give it. On the 24GB machine, that was two to five seconds. On the 16GB machine, it was 12 to 15 seconds. The 24GB also generated a few more tokens per second, though that figure ended up pretty close. It was really the time to first token that separated them.
A few other tasks leaned the same way. Compiling code in Xcode took 2:48 on the 16GB and 2:26 on the 24GB. The Blender Classroom scene took 48.63 seconds on the 16GB and 37.16 seconds on the 24GB. The one that surprised me was a Lightroom Classic export of 500 RAW photos to JPEG: about 3 minutes 18 seconds on the 16GB and 1 minute 54 seconds on the 24GB. Adobe apps seem to lean on RAM more than most, so depending on the apps you use, you can notice a real difference.
Disk speed and benchmarks
Plenty of things were basically the same. Disk speed was very close in Blackmagic Disk Speed Test (6,409 / 6,400MB/s on the 16GB, 6,630 / 6,511MB/s on the 24GB), and compressing large files took the same time on both. In the synthetic benchmarks, Geekbench 6 single-core was 4,180 vs 4,334 and multi-core was 16,736 vs 17,874. The GPU scores were basically identical, 75,678 vs 75,874 in Geekbench 6 Metal. Cinebench 2024 multi-core moved more, 928 vs 1,152.
Chip differences
Both machines have the same M5 chip with a 10-core CPU and a 10-core GPU, so on paper RAM is the only difference. But no two chips are exactly identical, and a few of these results can’t really be explained by memory. Cinebench 2024 multi-core is a CPU test that barely touches RAM, yet the 24GB scored 1,152 vs 928 on the 16GB. Geekbench 6 multi-core was about 7% higher on the 24GB too, while the GPU scores were identical.
Small chip-to-chip differences, how warm each machine was when a test started, and background tasks after setup can all move a CPU score by a few percent. So I would treat the gaps in pure CPU benchmarks as normal variation, not a RAM win.
For some context, I later tested another 16GB M5 MacBook Pro for my 2026 comparisons, on newer software. That one scored 17,928 in Geekbench 6 multi-core, higher than this 24GB unit, rendered the Blender Classroom scene in 29.76 seconds, and exported the 500 photos in Lightroom in 2:23. So test round and software versions matter as much as the chip, and it’s best to compare numbers from the same round of testing.
What holds up is the memory-heavy stuff. Loading the AI model, time to first token, how many tabs it takes to hit memory pressure, and the Lightroom export all favored the 24GB by a wide margin, and the 24GB was still faster in Lightroom than that later 16GB unit (1:54 vs 2:23). That is where the extra RAM is doing the work.
Is 24GB worth it?
It costs about $200 more to take the MacBook Pro from 16GB to 24GB, and I think it is worth it. The RAM is not user serviceable, so you cannot add it later. Buying up front is the only shot you get. The 24GB gives you a bigger buffer and keeps memory from filling up as fast with modern apps or heavy multitasking, and that should also give you a couple more years of useful life out of the machine.
All in all, I was still pretty impressed with how well the 16GB held up. There is a reason Apple is making 16GB the baseline for its modern Macs. But the 24GB gives you that extra headroom for now and for the future, as applications keep getting more memory intensive and as more AI features arrive. If you want to save a little more, B&H is an Apple authorized retailer that often has strong sale prices, and the Payboo card can knock the cost down further.
The bottom line
The 16GB M5 MacBook Pro is genuinely capable, and for everyday work you may never feel a wall. But the 24GB model reaches memory pressure later, chews through local AI far faster, and handles memory-hungry apps like Lightroom with real gains. For roughly $200 on a part you can never upgrade, the 24GB is the configuration I would buy for peace of mind and a longer useful life. If you are running AI models, editing video, or doing 3D work, do not think twice about it.
