Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MakaniKai

macrumors regular
Original poster
Jul 26, 2023
194
247
I am considering getting an AVP but I know that M5 is rumored to be coming "some time in 2025." Personally, though, I have not heard of anyone complaining about CPU bottlenecking on the AVP. Would an M5 upgrade even be noticeable? Is it worth the wait?
 
I suspect the biggest differences will be:
- improved battery life from improved node and hardware optimizations (eg hardware ray tracing)
- AI capabilities, but I suspect it will be mostly just bringing it up to par with iOS
- more expansive virtual desktop support

And there might be a benefit for gaming with improved GPU and rumored support for Sony’s PSVR2 hand controllers.

Whether that’s worth waiting for or not really depends on your use cases.
 
I’ve never once thought about speed or the CPU while using Vision Pro. I know there have been discussions about Apple Intelligence on the current M2 Vision Pro. Personally I can’t imagine why whatever is current processing can’t be paused while a quick Siri request is processed. I’m hoping for AI in visionOS 3.
 
That’s what I thought too. My M2 MacBook can do AI stuff

Yeah you’d think an M2 could handle it since Apple has said you need an M series processor in Macs and iPads along with 8 GB of RAM to run Apple Intelligence and we know it’ll run on the A17 and higher in the iPhone but it was John Gruber, I think, who has said that because the AVP so heavily taxes its M2 that Apple needs something faster before the device can run AI.

My hope is that it was a premature or inaccurate statement on the part of whoever told Gruber that at Apple and that behind the scenes Apple’s hardware and software engineers on the AVP team are busy actively working on refining visionOS and refining / tweaking Apple Intelligence so that some or all of the Siri improvements can be run on the device without requiring new hardware.

I’m most interested in the refinements to Siri that aren’t even out for public consumption yet but that are said to be coming in iOS 18.4. IMO the virtual keyboard is a problem for the device. Trying to type words into that keyboard is a total PITA which makes it almost impossible to use for content creation when writing. Right now I’m using a hardware based keyboard and track pad but it would be great to have Siri for dictation as another option. If they can give us the latest changes to Siri, I don’t need anything else that I’ve seen so far in Apple Intelligence on the AVP and I’d happily pass on any M5 refresh.

If they truly can’t, not can but won’t, but straight up can’t then I might look at selling mine for as much as I can and biting the bullet on a model with an M5 but they’d have to improve it enough that they can bring all of Apple Intelligence to it including the improved LLM based Siri coming in iOS 19. Even then it would be a hard sell unless they can get more apps and games or immersive content out there.

There isn’t enough of that kind of content available out there yet to justify spending another grand or two on top of what I can sell mine for to buy refreshed hardware just for Apple Intelligence alone.
 
  • Like
Reactions: MakaniKai
I am considering getting an AVP but I know that M5 is rumored to be coming "some time in 2025." Personally, though, I have not heard of anyone complaining about CPU bottlenecking on the AVP. Would an M5 upgrade even be noticeable? Is it worth the wait?
AVP and visionOS are altogether unique and we don't have a history of releases to guess from.

However, since Apple has significantly improved specs for Macs and iPads to be able to run AI, I think any future AVP release will focus heavily on some kind of implementation of Apple Intelligence.

Doing a little speculative, hopeful hype, I'd suggest that advanced AI that can assist things like (peripheral-free) hand gestures, like typing, scrolling, pinching, etc., could actually "save" AVP and turn it into a must-have if done well.

I think AI could potentially enhance most aspects of AVP/visionOS and make it far more like the Minority Report fantasy that so many hope AR/VR would become.

The big bottleneck here is that Apple has to do most of it "on-device" and in a device that has all the components inside the headset.

And Apple burying AVP for good is also very much a possibility.

I lean more towards them taking the losses and tying to perfect it over 1-3 additional iterations. Who knows.

What I definitely can't see is a new model dropping in 2025 or 2026 -AVP launched only about 11 months ago, in early February 2024. 2025 is too soon.

My instinct tells me Apple right now is dedicated to making Apple Intelligence so good that most iPhone and Mac owners will want to upgrade to get it.

Perfecting AVP at the same time seems like too much at once, and maybe is downright counterproductive if Apple isn't completely sure were Apple Intelligence is in 1-2 years.

Just get an AVP now if you like it. You'll wait years and years for a future model, one that may never launch.
 
If there are rumors about it, and in general these supply rumors seem to be at least close, I wouldn't get one especially if that means you essentially leap 3 generation into the M line, and also considering you are paying $4k+ for the thing I'd want to get the latest and greatest. Usually any 1st gen Apple products go obsolete very quickly, and considering that the latest A series chips are approaching M2 speeds, I'd say that this product wouldn't be any different. See Apple Watch Series "0", iPhone 2G for examples of going obsolete quickly.
 
  • Like
Reactions: MakaniKai
The mesh shading and ray tracing of the M3, the massive improvement in single and multi core CPU speeds of M4, and whatever the R2 brings to the table will all make sizeable contributions to improving the VisionPro. We’ll have to wait to see what M5 brings to the table in general (actual GPU improvements? New tech?) but it will obviously include everything M3 and M4 already do.

I would also not dismiss the role of the neural engine, particularly for how the device generates spatial photos, etc.

All of those pieces of tech will take the present screens in the VisionPro to their limits… building towards higher quality screens (better viewing angles, faster refresh rates, etc) in a future generation.
 
Not true. The minimum requirement is M1, which AVP has covered. All indications are that AI will be implemented in VisionOS 3.
The Apple Vision Pro is a very different system with much different system overhead then something like a MacBook. As such an M2 processor, most likely gets used much differently.
 
  • Like
Reactions: MakaniKai
I’m in the camp of people thinking that Apple Intelligence can run on the current one just fine, they just haven’t implemented it yet because… Well look at visionOS, it’s barely finished.
Most of the default apps are still iPad clients, that’s certain to be a priority in 3.0.
Tons of apples default apps are just completely missing, those are probably coming eventually and certainly don’t need new hardware.
When the thing first released, it was missing critical iOS features like iMessage contact key verification and a proper Safari video viewer.
Even now you can’t even create folders on the home screen.

Apple Intelligence obviously had to be announced mainly for the iPhone, the iPhone is Apple‘s biggest money maker so of course it’s going to get it. The Mac is obviously going to get it, and the iPad just follows that.

but visionOS will eventually get it, probably in 3.0.
It’s probably near the bottom of the list of priorities though.
 
From my point of view, thinking about what it could bring in terms of what people usually consider as AI (mainly LLMs in the form of chatbots, or similar) is very boring/uninspiring. We already have that in other platforms, there’s nothing groundbreaking about bringing Apple Intelligence to the VP. They’ll do it, but it’s a minor point, it will not fundamentally change its UX. The VP already uses a lot of ML, for much more complex use cases.

I hope (and guess) they’ll focus on using performance in order to improve the core VP experience: making it seem like you’re really looking at the real world, blended with VR. That should be the top priority.
 
  • Like
Reactions: MakaniKai
I am considering getting an AVP but I know that M5 is rumored to be coming "some time in 2025." Personally, though, I have not heard of anyone complaining about CPU bottlenecking on the AVP. Would an M5 upgrade even be noticeable? Is it worth the wait?
AVP limitations revolve around it being v1: innovative hardware but with alpha OS, now beta with visionOS 2. Since M5 will probably come concurrent with hardware and OS improvements, certainly an M5 version will likely show significant upgrade. But not [IMO] for reasons of M5, but rather for all the other evolutionary improvements.

No one knows if "Is it worth the wait?" because we are all clueless as to what improvements may present, or how v1/v2 pricing may evolve.

The good news is that if I am correct then many visionOS 3 improvements may fully populate on existing AVP v1 hardware.
 
Last edited:
AVP and visionOS are altogether unique and we don't have a history of releases to guess from.

However, since Apple has significantly improved specs for Macs and iPads to be able to run AI, I think any future AVP release will focus heavily on some kind of implementation of Apple Intelligence.

Doing a little speculative, hopeful hype, I'd suggest that advanced AI that can assist things like (peripheral-free) hand gestures, like typing, scrolling, pinching, etc., could actually "save" AVP and turn it into a must-have if done well.

I think AI could potentially enhance most aspects of AVP/visionOS and make it far more like the Minority Report fantasy that so many hope AR/VR would become.

The big bottleneck here is that Apple has to do most of it "on-device" and in a device that has all the components inside the headset.

And Apple burying AVP for good is also very much a possibility.

I lean more towards them taking the losses and tying to perfect it over 1-3 additional iterations. Who knows.

What I definitely can't see is a new model dropping in 2025 or 2026 -AVP launched only about 11 months ago, in early February 2024. 2025 is too soon.

My instinct tells me Apple right now is dedicated to making Apple Intelligence so good that most iPhone and Mac owners will want to upgrade to get it.

Perfecting AVP at the same time seems like too much at once, and maybe is downright counterproductive if Apple isn't completely sure were Apple Intelligence is in 1-2 years.

Just get an AVP now if you like it. You'll wait years and years for a future model, one that may never launch.
I liked on your post, but I disagree with the statement "Apple burying AVP for good is also very much a possibility." IMO AVP represents an important new field of endeavor that will not be going away.

I also disagree with the idea that "Perfecting AVP at the same time [as AI] seems like too much at once." Apple is a $4T company than can and should easily direct huge resources into different directions concurrently.
 
The mesh shading and ray tracing of the M3, the massive improvement in single and multi core CPU speeds of M4, and whatever the R2 brings to the table will all make sizeable contributions to improving the VisionPro. We’ll have to wait to see what M5 brings to the table in general (actual GPU improvements? New tech?) but it will obviously include everything M3 and M4 already do.

I would also not dismiss the role of the neural engine, particularly for how the device generates spatial photos, etc.

All of those pieces of tech will take the present screens in the VisionPro to their limits… building towards higher quality screens (better viewing angles, faster refresh rates, etc) in a future generation.
Absolutely accurate. However the AVP's current limitations revolve around neither speed/ray tracing/CPU nor display quality. The AVP's current limitations are visionOS competence. Much like with iOS, improving the OS is the bottleneck.
 
The Apple Vision Pro is a very different system with much different system overhead then something like a MacBook. As such an M2 processor, most likely gets used much differently.

Yes, but the AVP has another processor that handles a lot of that system overload, something Macs lack.

The M2 is not the M1. And I find it hard to believe that my four year old iMac has the processing power to handle Apple AI, while the AVP does not.
 
  • Like
Reactions: MakaniKai
Yes, but the AVP has another processor that handles a lot of that system overload, something Macs lack.

The M2 is not the M1. And I find it hard to believe that my four year old iMac has the processing power to handle Apple AI, while the AVP does not.

Vision Pro is different in terms of latency requirements and thermal envelope. Most people are probably fine with a hiccup on Mac but less so on VP.
 
Vision Pro is different in terms of latency requirements and thermal envelope. Most people are probably fine with a hiccup on Mac but less so on VP.
Agreed. No doubt the inefficiencies of the OS is holding it back atm. I expect those bottlenecks to get better is future software generations.

As mentioned above, the Apple group devoted to implementing Apple AI didn't have an AVP to work with when system was being developed. Integrating that complexity into a completely new piece of hardware no doubt requires a lot more time.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.