M4 MacBook Pro Models Expected to Launch in Late 2024

[chucker23n1]

If, like yours truly, some users don´t need portability, one could save a buck or two going Mini/Mini Pro/Studio covering the required performance, or get more perfomance for the cash to be paid when the required spec/performance is portable.

No point spending on portability if it´s not needed. I got the pherperials I need, and it is beneficial with different upgrading cycles too.

Due to the mix of M2/M3 and upcoming M4 the direct comparison is somewhat lost, and if things develops to my prediction, there may be more changes to M4 in terms of core count, balance and so on. Would be fair to assume their AI approach will have notable impact. One cannot assume they just will follow the historical development be it for the chips or stuff like USB. Hopefully, they drop USB A entirely.

I'm not sure what you're arguing here.

I get the frustration that the Mac Studio isn't on the latest and greatest tech, but it isn't that old.

And increasingly, businesses simply prefer laptops. They're cheaper in terms of power consumption, and it's easier to quickly take one to a meeting or a coworker's desk.

• MacBook Pro has always been their biggest selling Mac.

Source?

(No, CIRP doesn't count.)

I imagine the Air outsells the Pro 2:1.

 

[theluggage]

But that makes "more MBPs than MBAs" even weirder!

...but even the "MBP" range starts with the $1600 M3 MBP which still appeals to consumers and is likely the second largest seller to the Air. Combine that with by-value rather than by-unit sales and it would make sense.

 

[hoodlum90]

This is interesting. The Macrumors article clearly states that the MBP is not expected to get OLED before 2026 and yet it seems the entire premise for the new machines coming this year is an increase in OLED shipments. What gives?

That is why it is always best to look at the original source of the rumor. Based on the link I provided, there is a definite bump in the supply chain this fall for the OLED panels at the size for tablet/Macbooks. Apple has secured most of those units.

It is interesting that Apple Insider actually captured this OLED inclusion in their rumor article heading for the same Macbook Pro tweet from Ross Young. For some reason Macrumors completely missed this. The source of the tweet should have been a hint to Macrumors as Ross only provides these updates when there is a notable change in the supply chain, not just a device refresh.

 

[Ronlap]

The base M4 MacBook Pro will probably have 8GB of RAM.

And a 256GB SSD. I really don't need a fully-loaded MBP but I have no choice because as a consultant, I have each client as its own UserID to keep them separated. And each needs dozens of offline files - which take up over a TB of SSD. Yes, I suppose I could use an external drive, but that's a pain.

 

[Allen_Wentz]

Timing makes sense to release in the 2 weeks or so after MacOS Sequoia is released.

I am wondering whether the RAM definitions will start to change a bit. We know that the M4 iPad has 12 GB, but only reports 8 GB. I've also seen that the Apple LLM takes up 4 GB, so putting 2 & 2 together and getting 5, I'm wondering whether 4 GB will be reserved for Apple Intelligence features and the remaining 8 GB will be used for regular stuff.

It would not surprise me if the current 8 GB models that support Apple Intelligence (M1-M3, A17 Pro) devote half their RAM to the LLM and have to fall back on using 4 GB for regular system tasks (and thus the performance is pretty poor when using Apple Intelligence). Then all new devices (M4, A18Pro etc) come with 12 GB so the devices have 8 GB for system memory even when the LLM is chewing through 4 GB...

My guess is that the concept of memory being reserved will not really apply, and is contrary to Apple's Unified Memory Architecture (UMA). Folks will no doubt continue to think in terms of certain functions requiring x amount of RAM, but operationally at processor level UMA means it is not that simplistic.

 

[PaperMag]

Source?

(No, CIRP doesn't count.)

I imagine the Air outsells the Pro 2:1.

So CIRP's data isn't valid, but your imagination is?

No thanks.

If you listen to podcasts and read blog posts by Mac pundits (the people who have inside info) for the last two decades, despite Apple never publishing sales numbers per model, it's been an open secret that MacBook Pro are Apple's #1 seller, especially because nobody buys Airs in enterprise. You can confirm this by working literally anywhere.

Maybe the $999 MacBook Air outsells the $1999 MacBook Pro by 2:1—I can't say—but historically the 13-inch MacBook Pro started at $1299—and those models made up the majority of MacBook Pros sold. I think you may be forgetting about those.

 

[Allen_Wentz]

It might be a good time for Apple's chip team to slow down releases of new chips.

Let the entire line up catch up, take a breath, before starting the process again.

No one benefits from yearly processor upgrades on desktop-grade chips for devices that last 3-5 years at least.
I'm not advocating forgetting about upgrades as happened a while back.
But doing so too regularly makes people go "well they upgraded 6 months ago, so i'll just wait another 6 months rather than buy now". If I dont upgrade my phone in the first 3 months, I wont in that 12 month cycle. Sales graphs show huge interest at release date and then massive tapering off. We dont need Mac and iPads to do the same.

We agree. However there is a lot more to producing tech devices than just the chip tech. There are issues around motivating workers, marketing planning, fiscal planning, etc. that ameliorate on favor of an annual cycle. Following a non-annual cycle or (worse) no planned cycle at all has huge implications in those other areas of the business of getting products out the door.

 

[PaperMag]

Yeah, but then the iMac number suddenly becomes hard to believe. Four times as much revenue from the iMac ($1,299 to $2,699, and a mid-range seller) than from the Mac Studio ($1,999 to $8,799, and probably not that many fewer sales to make up for the four times)?

iMac has always been's Apple's most popular desktop. They've said it themselves in keynotes.

 

[deconstruct60]

Gurman said in April that the Mac Pro is "set to get the new Hidra chip,” a “top-end” version of the M4. And RAM would supposedly go from 192 GB to a 512 GB option with that chip.

I'm calling that the Extreme chip, but I'm obviously speculating.

512GB is doubtful (at least with a single die). Apple would need to reverse course on ECC support and they likely are not. Even if they did the capacity would trend down a bit because need to keep track of the checksums.

If Apple went from 192GB to 256GB , that would be a useful increase. ( a bit dubious with no ECC). A Nvidia's upcoming Blackwell B200 has 192GB HBM memory (with ECC). They'd still gap them with 256GB. 512GB is off chasing DDR5 modules which I doubt Apple is going to go down that path.

EDIT: Learning more, Hidra is code for the M4 Ultra. But it will be its own chip (not two M4 Maxs put together with an interposer, like the M2 Ultra was). For instance, the M4 Ultra won't have any efficiency cores, only performance cores. The rumor is that Apple will combine two M4 Ultras with an interposer to form an M4 Extreme. If that's the case, we'll learn more the closer we get to WWDC 2025.

That seems unlikely. If Apple was going to huge effort to dump 3D packaging for the Ultra, why would they go back to even more expensive 3D packaging? If it was 'pain' point they were trying to avoid, why go back to it ?
Going to a die that is as big as possible would not be a way to save money unless the alternative was actually more expensive. The notion that they are going to 'radically cheaper' interposer options would throw bandwidth and latency substantially backwards.

The other problem going to have if that a monolithic Ultra die would get more square shaped to fill up the whole reticle limit. That would impact how the memory packages are placed, which ripples into placement of the two packages. ( there are folks who imagine that this Hidra means abandoning LPDDRn which seems likely wishful thinking. )


Dumping E cores isn't going to save much die space if about to bust the reticle limit. It isn't going to buy significantly more P cores. The M3 Max keeps the same 4 E cores as the M2 (while the M3 Pro went up to 6). Pretty likely still have 4 E cores left on a monolithic die. P core count probably just 'doubles' from Max. The rest of the 4 E core count would go toward trying to limbo under the reticle limit. M2 Max was a bigger die than M1 Max. The M3 Max got bigger again. TSMC N3E backslides on SRAM/cache density to TSMC N5 ( M2 Max) levels. The 'bloat' that popped up at M2 Max is likely just as much there at M4 ( if not more. If the display controllers and GPU's and NPUs get cache upgrades).


Folks seem to want Apple to create some Nvidia Datacenter B100/B200 'killer' package (or AMD MI300 killer ) and it probably is not going to be that. Likely not some AMD EPYC (or Xeon 6 datacenter ) 'killer' either. Nor does Apple need to chase the Nvidia x090. As long as the "AI mania" is pushing Nvidia, AMD, and others to use up all available 3D packaging capacity, it would make sense for Apple to take the M-series off of that pragmatically limited resource. ( 'Apple has got money they'll just outbid them for the scare resource'. Apple has money because they generally do not get into biddings wars. They don't try to crank their costs as high as possible. )

 

[chucker23n1]

My guess is that the concept of memory being reserved will not really apply, and is contrary to Apple's Unified Memory Architecture (UMA). Folks will no doubt continue to think in terms of certain functions requiring x amount of RAM, but operationally at processor level UMA means it is not that simplistic.

UMA doesn't change anything about how memory is allocated.

 

[deconstruct60]

iMac has always been's Apple's most popular desktop. They've said it themselves in keynotes.

They have not said that since the M-series transition. It was true when Apple kneecapped the Mini and cranked the Mac Pro entry incrementally higher on each release. But that was Apple lowering the fratricide in the desktop line up.

They aren't doing that anymore. The iMac has exactly the same processor as MBA 13/15 and Mini. Apple has said the MBA 13" is their best selling product. Recently said the MBA 15" is the best selling 15" laptop model (partially because the Windows 15" market is so heavily fragmented with a couple more orders magnitude more models to choose from.)

Apple replaced the 27" iMac with the Mac Studio. That isn't going to increase iMac aggregate sales. The comatose 24" M1 iMac extremely likely didn't expand unit sales to cover that lost ground.

 

[Allen_Wentz]

That reminds me that I made most of my work for college on my C128 with 1.5 megabyte memory expansion for GEOS128. That was enough memory back then. Visually the printed results didn’t look much different than a black&white Pages document made on a 16GB Mac today.
😎

Being inured to what we paid for one MB of memory back then (and to the obvious impact the extra RAM had on operation) may be why today I do not object to paying $800 for 64 extra GB of RAM.

 

[dynamojoe]

And increasingly, businesses simply prefer laptops. They're cheaper in terms of power consumption, and it's easier to quickly take one to a meeting or a coworker's desk.

There's also something to be said for a built-in five hour UPS. Much less likely to lose your work to a brownout.

 

[Allen_Wentz]

So the Mac Mini will be more powerful then the Studio for several months. I don’t buy that, not for a second.

The folks who buy Studios and consider Minis realize the differences between the two that transcend the timing of chip versions. Except for a few with time-immediate needs, the (generally sophisticated) buyers in the market for Studios would not be wowed into buying a lesser level chip, less ports, etc. in a Mini just because the number after the M is (briefly) newer.

 

[deconstruct60]

It’s really disappointing to see Mac Studio and Mac Pro users consistently having to wait an excessive amount of time for updates. Apple, I understand that Mac Pro users might expect longer intervals between updates, but the Mac Studio should be updated as often as the MacBook Pros. It’s frustrating that the Mac Studio often gets updated last, only for the next generation of chips to be released within six months. It feels unfair to spend a significant amount of money on technology that quickly becomes outdated.

Outdated how? If there is no released M4 Ultra the M2 Ultra is still top of the line until replaced.

The issue is the the Mac Studio is coupled to the Mac Pro via the Mn Ultra. If Apple wasn't so OCD they might be able to update just 'half' of the Studio line up. But likely they just don't want to.

The M2 Max/Ultra don't get slower if something new gets release. If the M2 Max met production requirements to get work done; it is still working. At the higher system price level there are few folks dumping systems on a 12 month cycle rate.

Apple relatively quickly dumping the M2 Pro/Max out of the MBP 14/16 probably has a ripple effect of making the Mini/Studio which also pick up those SoCs move slower. Quickly dump the M3 Pro/Max and similar 'no free lunch' on cost recovery hiccups there also (e.g., wait until cheaper to move to M4 generation so can use some of that money to cover short cycle on M3 gen. )

While Apple churns at the iPhone yearly they don't stop selling those iPhones for 3 years. ( it trickles down the product line and sold as n-1 and then n-2 year old model). The SoC is also is leveraged as a 'hand me down' SoC to entry iPads , AppleTV , etc. They keep selling the SoC of more than several years. The Mn Max and Ultra disappearing every 12 months is highly likely not economical.

( But apple is going to toss some into Private Cloud Compute (PCC). That is not going to be a bottomless pit. The initial ramp will consume a substantial amount, but those servers are not going to get dumped on a 12 month basis either. Even more so for a 'free' (not directly paid for) service. It will be years to get cost recovery on those systems with no direct payments. )

The volume of SoC sold as they get more expensive goes down.

At some point the MBP 12 month schedule probably will slow down. I suspect Apple is trying to minimize the dynamic changes happening in the Windows markets right now in laptops. The SoC competition there is far higher than it has been over a decade. AMD is executing (Strix Point/Halo appear quite good) and Qualcomm has something more than decent. Intel is working much harder to keep up.

Intermediate term Apple takes a hit on desktop to keep their foot on the 'gas pedal' for MBP until Windows get more clear stable path. But they can't keep robbing Peter to pay Paul forever. Either that or they need more consumer products to push the older 'desktop' processors to.

 

[aknabi]

after 3 years we need a re design, and get ride of the huge notch for just a web cam

They didn't add a notch... they removed 90% of the top bezel... I for one don't want the big bezel back.

 

[Ethosik]

Outdated how? If there is no released M4 Ultra the M2 Ultra is still top of the line until replaced.

But it is not top of the line that is the point. Many workflows can take advantage of the significant GPU improvements that the M3 has made. Thus, for those workflows, the M3 Max is top of the line. That is the frustration here. Mobile chips beat out last years top of the line desktops that are double the cost of the laptop.

 

[Allen_Wentz]

Precisely. Mac Studio is less relevant today, and it will be two generations old if this rumor and the other one is true we won't see Mac Studio get updated until mid-2025. Apple has been frustrating pros since the trash can Mac Pro. They need to focus more on it.

Agreed Apple should give better attention to the Studio and to the MP. However I do not think "Mac Studio is less relevant today." I think the Studios are important boxes for Apple's core ;~) desktop users. Reality is that one or the other of the M2 Studios with max RAM beautifully served Apple's core desktop users - - and continues to do so. IMO Studio upgrades should be more routine, but the M2s are so ideal that perhaps Apple just decided M3 incremental improvements were not worth the effort.

P.S. I am exactly in that core market for Studios.

 

[WRONG]

Can’t come soon enough, desperately need an M4. But in meantime my core solo Mac mini from 2005/2006 will serve my workflow needs.

😜

If that's true, utter respect for your solo Mini! 😆

 

[rp2011]

This is too much for me. I am looking at the 14” M3 Max from the refurb store. I just can’t bring myself to spending 3 grand plus on a laptop. I was looking at the Studio but can’t wait till next year for an M4 update if it is next year. And the M3 is quite a bit more powerful then the M2 Max.

Yep. I'm on the same boat buddy, and I'm on an M1 Max Studio. Best desktop I have ever owned.
I also have a newer Windows machine with Nvidia GPU for rendering and it's always a headache.

 

[tkrsh]

no oled = wont upgrade

 

[chucker23n1]

Outdated how? If there is no released M4 Ultra the M2 Ultra is still top of the line until replaced.

The issue is the the Mac Studio is coupled to the Mac Pro via the Mn Ultra. If Apple wasn't so OCD they might be able to update just 'half' of the Studio line up. But likely they just don't want to.

I don’t think the MP is the issue. I just think they didn’t bother making higher-end variants of the M3, because volume isn’t that high and the M4 was just around the corner.

 

[Allen_Wentz]

512GB is doubtful (at least with a single die). Apple would need to reverse course on ECC support and they likely are not. Even if they did the capacity would trend down a bit because need to keep track of the checksums.

If Apple went from 192GB to 256GB , that would be a useful increase. ( a bit dubious with no ECC). A Nvidia's upcoming Blackwell B200 has 192GB HBM memory (with ECC). They'd still gap them with 256GB. 512GB is off chasing DDR5 modules which I doubt Apple is going to go down that path.




That seems unlikely. If Apple was going to huge effort to dump 3D packaging for the Ultra, why would they go back to even more expensive 3D packaging? If it was 'pain' point they were trying to avoid, why go back to it ?
Going to a die that is as big as possible would not be a way to save money unless the alternative was actually more expensive. The notion that they are going to 'radically cheaper' interposer options would throw bandwidth and latency substantially backwards.

The other problem going to have if that a monolithic Ultra die would get more square shaped to fill up the whole reticle limit. That would impact how the memory packages are placed, which ripples into placement of the two packages. ( there are folks who imagine that this Hidra means abandoning LPDDRn which seems likely wishful thinking. )


Dumping E cores isn't going to save much die space if about to bust the reticle limit. It isn't going to buy significantly more P cores. The M3 Max keeps the same 4 E cores as the M2 (while the M3 Pro went up to 6). Pretty likely still have 4 E cores left on a monolithic die. P core count probably just 'doubles' from Max. The rest of the 4 E core count would go toward trying to limbo under the reticle limit. M2 Max was a bigger die than M1 Max. The M3 Max got bigger again. TSMC N3E backslides on SRAM/cache density to TSMC N5 ( M2 Max) levels. The 'bloat' that popped up at M2 Max is likely just as much there at M4 ( if not more. If the display controllers and GPU's and NPUs get cache upgrades).


Folks seem to want Apple to create some Nvidia Datacenter B100/B200 'killer' package (or AMD MI300 killer ) and it probably is not going to be that. Likely not some AMD EPYC (or Xeon 6 datacenter ) 'killer' either. Nor does Apple need to chase the Nvidia x090. As long as the "AI mania" is pushing Nvidia, AMD, and others to use up all available 3D packaging capacity, it would make sense for Apple to take the M-series off of that pragmatically limited resource. ( 'Apple has got money they'll just outbid them for the scare resource'. Apple has money because they generally do not get into biddings wars. They don't try to crank their costs as high as possible. )

"trying to limbo under the reticle limit." was the best internet verbiage today

 

[deconstruct60]

But it is not top of the line that is the point. Many workflows can take advantage of the significant GPU improvements that the M3 has made. Thus, for those workflows, the M3 Max is top of the line. That is the frustration here. Mobile chips beat out last years top of the line desktops that are double the cost of the laptop.

That is a narrow case. You are pointing at a subset of GPU loads. The 'top of the line' doesn't have to be better at everything possible. Back in the days where Mac Pro had modular GPUs folks could throw a newer GPU into a old Mac Pro and beat the entry GPU configuration spec of the new Mac Pro model. If want to cherry pick corner cases Apple had a chart where the M1 Max was doing better than a 3090 . Broad picture workload wise that really didn't hold up.
There are many workloads where M2 Ultra still comes out on top.

Workloads where the thermal system of the MBP 14" keeps up with the thermal system of the Mac Studio.

Apple isn't really building "mobile" and 'desktop' chips. It is just one. If someone bought a studio to run a large mix of single threaded stuff then the next iteration out will beat it. ( nothing to do with mobile or not). Apple single threaded is pragmatically uniform through the whole line up. There is no overclocking heat monster in the desktop line up where throw perf/Watt efficiency out the window just to 'win' ST crown.

The vast majority of users who spend higher end Mac Studio or Mac Pro money don't churn their systems on a 12-16 month basis. They buy systems to work for several years. If the system 'underperformed' when it was new then probably should not have bought it. It is not a bragging rights contest device , it is aimed at being a 'get work done' device. The work being done before made money, then it still is. Paydown what you have and buy something new when it is paid off (while paying off the rest of bills suppose to also pay).

There is not enough Ultra volume to kill of these SoCs quickly. Ultras following a n+2 pattern likely would be closer to economical than trying to flush the very large die down the drain every 12 months. The die/package is just too big to be thrown away that quickly without cranking the purchase cost substantively higher. ( which would only get Apple into a pricing death spiral. Higher Ultra means fewer buyers. Fewer buyers means higher amortization costs. rinse-repeat). To get enough customers to amortize the costs over they need to sell it longer.
The fewer Apple products units shipped the longer the SoC likely stays around waiting on an upgrade.

[ This is the 'no free lunch' that happens when Apple completely decoupled from the Windows desktop market (which is bigger than Apple's whole share) and from the even larger server x86-64 market. The Unit volume goes way down, which means they are not going to be able to match the pace. ]

 

[Allen_Wentz]

UMA doesn't change anything about how memory is allocated.

We disagree. UMA means memory is in one place, accessed as needed rather than being copied back and forth or reserved for different operations. A semantic difference perhaps, but IMO it seems significant.