M3 Max Now or Wait for the M4 Max?

[lionfleet]

I recently bought a 16" M3 Max Base Model (14 core CPU, 30 core GPU)...

I was planning to upgrade from my M1 Max (10 core CPU, 32 core GPU) but I haven't opened the new computer yet as I've just been realizing that my current M1 Max is totally fine for my needs. I really didn't need to upgrade so soon. lol

I'm thinking about selling the unopened M3 Max on Facebook Marketplace and waiting another year for the M4 Max to come out. I got a deal on the M3 Max and could probably get my money back.

I've heard the M4 Max will have some new AI enhancements... although I'm not sure how much that will actually make a difference in real life?
So my question is... do you think it's worth selling and waiting for the M4 Max to get the AI enhancements?

Or should I just open the M3 Max and use it now?

My usual upgrade cycle is every 3 years to keep up with the latest tech but spread out every 3 years or so to maximize out of pocket expenses.

 

[dmccloud]

You are in a bit of a Catch-22 scenario here. Since any M4 information out there right now is mere speculation and rumor it's hard to say whether it will be a big upgrade from the M3 series. Once WWDC hits and we see the announcements, there would be a better answer to your question.

 

[Chuckeee]

Traditional wisdom is if your existing system is satisfactory then you should wait. There is always something better coming so wait if you can.

 

[TechnoMonk]

For your use case unless you are running out of RAM or system performance is low. Stay put on M1 Max and upgrade later. My M1 Max 16 MBP works great, I have 64 GB RAM, for my use cases M3 max is about 25% faster, but I could Definetly use 128 GB RAM. I am holding out till M5 or may consider M4 if it comes with 256 GB RAM.

 

[geta]

Simple - if you need new computer, the correct M3 Max will be nice upgrade, but if your correct MBP still do everything you need, wait for the new M4 Max.

 

[pksv]

For your use case unless you are running out of RAM or system is low. Stay put on M1 Max and upgrade later. My M1 Max 16 MBP works great, I have a 64 GB RAM, for my use cases M3 max is about 25% faster, but I could Definetly use 128 GB RAM. I am holding out till M5 or may consider M4 if it comes with 256 GB RAM.

I made a mistake getting 32GB RAM with a Max chip, it's running out of memory from time to time. If I had opted for 64GB I would not be looking at an upgrade any time soon.

 

[pksv]

Yep. I max out on RAM, stick to 1 TB SSD. I have external thunderbolt storage rack. RAM is more binary, either you have enough or not. I upgrade every 5 years for professional use.

Had a good deal on that machine and I reasoned that it would be a temporary machine, would wait for the new version, and then upgrade. Well, I was not wrong 😁

Whatever Apple or YouTubers want you to believe, 32GB RAM is NOT enough for professional workflow unless you want to open and close apps constantly. I would say 64GB is the minimum, also don't be fooled by 36 or 48 with M3 Max chips. On M3's OS uses more RAM than on M1 or M2 so you're left with less.

With M3 Max you should only choose 64, 96, or 128 variants.

 

[xraydoc]

I bought an M3 Max a few months ago. No regrets. Running 3x 5K displays... and it doesn't even break a sweat pushing over 44 million screen pixels.

 

[Jackbequickly]

I’m upgrading from an i9 2.3/64GB/2TB 16” to a 16/40 48GB 1TB 16” M3 Max.

I always purchase from the Apple refurb store so M3 time is now. I’m saving £620 on RRP. M4 time would be 6 months after launch.

Instead of waiting on the M4, I took advantage of Best Buy's deal and picked up my M3Max MBPro with 1tb and 36gb RAM. I know the M4 will be fast but the price on the M3 Max was awesome. It runs my 3D CAD modeling with ease!

It will be interesting to see how the M3 Max compares to the M4 but I could not be more happy!

 

[Mity]

Are you still within the return window? If so, return the M3 and wait. I'm kinda in the same position as you. My coding computer is a 2019 16" Intel MBP so upgrading to the M3 Max makes sense for me but if I could wait until June, I would. If you have an M1 Max, I definitely think you can wait.

Edit: What are the specs on the M3 Max and what price are you asking?

 

[MrGimper]

I’m upgrading from an i9 2.3/64GB/2TB 16” to a 16/40 48GB 1TB 16” M3 Max.

I always purchase from the Apple refurb store so M3 time is now. I’m saving £620 on RRP. M4 time would be 6 months after launch.

 

[TechnoMonk]

I made a mistake getting 32GB RAM with a Max chip, it's running out of memory from time to time. If I had opted for 64GB I would not be looking at an upgrade any time soon.

Yep. I max out on RAM, stick to 1 TB SSD. I have external thunderbolt storage rack. RAM is more binary, either you have enough or not. I upgrade every 5 years for professional use.

 

[coffeemilktea]

So my question is... do you think it's worth selling and waiting for the M4 Max to get the AI enhancements?

The M4's promises of "AI enhancements" are too vague to make any kind of decision over. However, if you're really into AI (especially generative AI), I can say the base model of the 16" M3 Max has too little RAM for heavy AI work.

I'd recommend waiting for the M4 and then getting more than the base RAM. (remember that on Apple Silicon, the GPU shares RAM with the rest of the system, and it's limited to using a maximum of 75% of the total RAM).

 

[TechnoMonk]

For those who want to track official Apple MLX/AI engineering on twitter, here is a good one.

https://twitter.com/x/status/1783615836063969700

 

[xraydoc]

Is it dead silent running 3x 5K displays?

Yep

 

[jlc1978]

Whatever Apple or YouTubers want you to believe, 32GB RAM is NOT enough for professional workflow unless you want to open and close apps constantly. I would say 64GB is the minimum, also don't be fooled by 36 or 48 with M3 Max chips. On M3's OS uses more RAM than on M1 or M2 so you're left with less.

With M3 Max you should only choose 64, 96, or 128 variants.

That depends on your workflow. There are plenty of use cases where 36GB of RAM are plenty; while others may require more.

 

[Lotuskid]

i can share my own personal experience - i had an m1 pro 16gb ram and 1tb storage. the storage was never an issue, but occasionally when i did some photo editing jobs in lightroom i would experience some slowdown. recently i had a ram usage warning as i had lightroom open, safari playing the nba game, and i might have had a final cut project open as well.

for me the ram i think was a factor. there would be occasional lag as well cycling through photos, saw the beach ball here and there.

i purchased an m3 max that i later returned in favor of an m1 max with 64gb of ram. the total cost to upgrade was about $800 dollars and for that i did get a better cpu than the m1 pro along with quadruple the ram

everyone's use case is different, but as of now i feel very good that this configuration will last several more years for my amateur photography and videography projects, and i get to see whats coming down the pipeline with m4 and future generation chips

 

[DaveEcc]

On Geekbench we can see iOS M2 and Mac M2 scores are quite similar, thus iOS scores and Mac scores are valid comparisons. There are no iOS M3s to compare to, but we can use the iOS M4 scores and compare to Mac M3.

	iOS M2	Mac M2	Mac M3	iOS M4
Single	2570	2578	3050	3668
Multi	9797	9657	11713	13276
Compute (CL/Metal)	/ 41267	25349 / 41465	30317 / 47303	/ 53202

All M4 Macs should see the same 20% single core improvement. We don't know the core configurations on M4 Pro or Max, but we can assume we'd get at least a similar 13% multi-core improvement, and 13% compute.

Can we hope for more? We can compare M2Max vs M3Max improvements, vs M2 and M3. M2Max to M3Max was 10%, 41%, 6.7%. While the multicore score jump was big, the M4 is a bigger leap for both Single core and Compute. The Base M2 to M3 multicore jump was "just" 21% compared to the 41% on the Max, so the limited improvement on the base M4 isn't necessarily an indication of what we'll see on the M4 Max.

If the M4Max follows the previous generation of doubling the base multicore improvement, that would be a nice 26% jump... far better than an incremental improvement.

It's all speculation, with too many assumptions and "if"s, to say anything for certain, but I don't see any reason to not expect M4 to be decent update, even if it doesn't quite give us a 41% multicore jump.

 

[TechnoMonk]

The M4's promises of "AI enhancements" are too vague to make any kind of decision over. However, if you're really into AI (especially generative AI), I can say the base model of the 16" M3 Max has too little RAM for heavy AI work.

I'd recommend waiting for the M4 and then getting more than the base RAM. (remember that on Apple Silicon, the GPU shares RAM with the rest of the system, and it's limited to using a maximum of 75% of the total RAM).

Bro Science alert! Are you saying 16GB on my 64 GB M1 Max is not accessible? I have a base M2 MBA which is running a llama3 8B model. My iPad Pro runs that model without any problem. Here is a snapshot of m M1 Max during a 10-thread, multi-worker GPU intensive Training with 173 GB memory. Sure, it was swapping heavy, but it had no problem running. Apple MLX team is very active on Twitter, and they just released a new framework for LLM on devices, which is even more efficient. Apple is hardly vague, they are publishing multiple papers, making constant improvements to MLX. The biggest gains weren't even the hardware improvements from M1-M3; it was on libraries that are optimized for Apple silicon.

M2 Ultra is still the king of Apple silicon when it comes to training and inference of LLM/Transformer models. Apple recently published an enhancement to Matrix multiplication with 25040% improvements. M4 may see some of these regular options using dedicated neural engine.

 

[Eric_WVGG]

I'm on the min-specced M3 Max, it's just unreal.

It really sounds like the M4 series is going to mostly be focused on the "neural engine" and all this ML/AI stuff, and I think your decision should mostly be based on what you think of that.

I'm personally a skeptic — mostly smoke and mirrors, will become useful someday but we're nowhere close. Additionally, this is all on-device stuff — it'll make Siri better (probably), and be great if you're a developer of AI stuff, but as a consumer… I just don't see it.

The M4 Max will probably be 10-15% faster due to the usual process improvements, and have substantially better AI that right now isn't used for anything practical. I don't think that's worth waiting on, personally.

 

[dmccloud]

I'd recommend waiting for the M4 and then getting more than the base RAM. (remember that on Apple Silicon, the GPU shares RAM with the rest of the system, and it's limited to using a maximum of 75% of the total RAM).

If you allowed the GPU to use 100% of the system RAM in those scenarios you would have nothing left to run the OS and everything outside the AI/ML operations. This is still a massive improvement over x86 systems with integrated graphics because those systems will partition the system RAM into what's usually a 75% CPU, 25% GPU ratio, and those are fixed partitions which the system sees as separate RAM pools.

 

[Churchman]

I am about to buy the M3 Max 48G-ram MBP 16 inch. (The plan is the buy it before the AI enhanced M4 chip comes out. I want to stay a thousand miles from AI chips and such, even though forthcoming OS's and their updates will be awash in AI functionality.)

I don't do the heavy lifting that requires 64G ram. But I do want a machine that is robust enough to last 10 years.

The M3 Max MBP-16 will replace my 11 year old 2013 MBP 15 inch with the Intel Quad core i7 chip and 16GB RAM & Intel Iris Pro 1536 GPU. This machine has served and still serves me well... battery is still quite strong... the fan (which used to never activate) comes on strong once every few days or so and then only for short bursts... 94% of SSD is free with most files stored externally... and it still looks pristine.

QUESTIONS - Is there any value to my current 2013 MBP Intel machine? Is it worth selling? Or is there anything about it that is upgradable at all aside from the battery, such that I can turn it into a dedicated desktop or a dedicated streaming device hooked up to the large flat screen TV?

Thanks!

 

[geta]

I am about to buy the M3 Max 48G-ram MBP 16 inch. (The plan is the buy it before the AI enhanced M4 chip comes out. I want to stay a thousand miles from AI chips and such, even though forthcoming OS's and their updates will be awash in AI functionality.)

I don't do the heavy lifting that requires 64G ram. But I do want a machine that is robust enough to last 10 years.

The M3 Max MBP-16 will replace my 11 year old 2013 MBP 15 inch with the Intel Quad core i7 chip and 16GB RAM & Intel Iris Pro 1536 GPU. This machine has served and still serves me well... battery is still quite strong... the fan (which used to never activate) comes on strong once every few days or so and then only for short bursts... 94% of SSD is free with most files stored externally... and it still looks pristine.

QUESTIONS - Is there any value to my current 2013 MBP Intel machine? Is it worth selling? Or is there anything about it that is upgradable at all aside from the battery, such that I can turn it into a dedicated desktop or a dedicated streaming device hooked up to the large flat screen TV?

Thanks!

You won't get anything for your old MBP... instead, if it's in working order, I'd give it to a friend or family member who in needs.

 

[Mac Hammer Fan]

Based on sources the M4 max will be approx. 20-25% faster than the M3 Max. But you have to wait until early 2025.
I have a M1 Mac Studio but I definiteley won't upgrade to a M3 Max, but perhaps to a M4 Max.

 

[steve123]

I'm personally a skeptic — mostly smoke and mirrors, will become useful someday but we're nowhere close.

The M4 Max will probably be 10-15% faster due to the usual process improvements, and have substantially better AI that right now isn't used for anything practical. I don't think that's worth waiting on, personally.

The recently announced upcoming versions of FCPX and Logic Pro have some impressive AI features now.