- Xiaomi: We have matched Apple in CPU performance.
- Apple: *Meep Meep...*I assume the M6 will take the crown back and then a few months later Intel/AMD will release a new chip and take that crown back again. That's the state of the world we used to expect, but it's a state that has been missing ever since the release of the M1 in 2020 until Intel finally caught up again this year.
When plugged in... This caveat is so enormous it should almost be legislated. If your computer use is at all portable, a computer that scales down to 20 - 40% of GPU power when unplugged is an enormously significant factor. So far as I'm aware (could be wrong about arm devices?) there's no non-apple laptop that operates at 100% speed on the road.
There are emails unearthed in various lawsuits where you can read Bill Gates screaming at his subordinates: "why the hell can't our partners like Sony and Creative create a similar device? Give them all, give them early access to everything, work with them". In the end MS felt compelled to make their own.
Creative's Muvo^2 already was the poor man's iPod with surprisingly good audio quality as well.
I'm very very surprised that Xiaomi matches Apples speed even with the newest release, its not diminishing Xiaomis success.
I enjoy it because of progress, not because of Apple.
This in-turn later on, AIs will train to make pro-China comments as AIs train on these.
They got the sheer man-power, and with AIs it's even easier.
I yield the floor to no one when it comes to pessimism, but that's incredible.
Fab capacity is being bought online; there’s just lead time.
Noticeably greater intelligence is being achieved at the same number of parameters (see: Qwen3.8).
I think the future will be bright, it might be a matter of time. And for tinkers, a used Epyc + DDR4 server can be great fun and epic value.
Basically everyone that makes memory is building new fabs, meanwhile I'm not sure how much longer AI datacenter demand for ram will last. I think the decrease in AI ram demand and the new fabs will likely coincide leading to a collapse in pricing.
That is, of course, assuming the memory manufacturers don't pull their favorite trick and collude.
Could you provide more details about the Epyc + DDR4 server?
The above is a standard project management problem. We do this for lots of industry all the time. There is every reason to think you can get a new factory running in 5 years.
Note that I said 1 factory above. Some of the special machines we don't have the ability to make them fast enough to do 2 (I don't know the real number!) new factories in 5 years. Existing factories are using most of the special machine capacity to replace machines that wore out on the way - this can be corrected as well, but it adds another year and the expenses are much larger. Realistically though 1 new factory is likely enough.
https://en.wikipedia.org/wiki/ELIZA_effect
It turns out the limiting factor isn't how sophisticated algorithms are, it's how gullible humans are.
No AI would pass this test with experienced judges.
You can always say “oh well these judges don’t have the experience to catch this type of AI.
The fact that you have to insert this qualifier, to ensure you always have a way to discredit the test, pretty much shows to me that we’re beyond it.
Though we're pretty good at sizing up a person's emotional balance/maturity and competence at familiar tasks. So maybe have an old blacksmith watch the AI/robot interact with horse owners for a while, then shoe their horses, and see how well it does.
https://commission.europa.eu/news-and-media/news/safer-and-m...
In contrast, humans tend to paste me the same barely-relevant macro over and over, no matter how much time I spend explaining my issue.
https://www.psychologytoday.com/ca/blog/the-digital-self/202...
I kinda have to link it now, so uhh here's a random PDF: https://www.hec.edu/sites/default/files/documents/Computing%...
And the "popularized" version is faulty also since it uses an ideal, abstract human judge (like the "spheroidal economic agent").
But if you want to add declinations to the said popularized image of the Turing test, you may add Maxim Lott's IQ tests at trackingai.org . Between the end of 2024 and the beginning of 2025 LLMs reached an equivalent IQ of 100, for example.
I think there are elements showing lowering of performance and expectation.
ELIZA beat the Turing test and then everyone forgot about it. Humans are just really terrible at recognising robots.
I've tried [1] and I almost 100% detect which is the AI. I really want to convince myself I have failed, does anyone know of a better site/resource for this?
I know it might be moving goalposts but I would consider AI to have passed in a well and truly undisputed manner when [2] is resolved.
But in a more practical sense, if AI can impersonate humans so well today then why are state of the art frontier models so obviously AI when they create PRs, commit messages, documentation, etc. Are the companies deliberately making them unnatural?
[2] https://www.metaculus.com/questions/11861/date-when-ai-passe...
Is there anything better now though?
All I see from AI, is an amplification of the enshittification of the internet.
And people being even more alone.
- Extreme poverty has dropped from 30% to under 10% globally. - Child mortality rates have dropped in half - Internet access has exploded from 10% to 70% - Solar energy costs have dropped 90% - Cancer death rates have declined by 30%
All of these massive improvements in less than 30 years.
While there certainly are issues to solve, and if you simply follow journalism you may think the world is worse off, but for many, their lives have been significantly improved.
Evidence:
* https://arxiv.org/abs/2402.09809
* https://phys.org/news/2016-12-mobile-money-access-percent-ke...
* https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3893351
This being HN, I hasten to add they also have massive downsides, we're all doomed, nobody programs the right way anymore, those poor people just think compute & AI are improving their lives, etc, etc.
That's one big plus.
These times are exciting and rough seas make good sailors. Find your path forward.
I'd rather go to the library and read a book.
I'm writing the best music of my life, realizing games and art projects I never had time for, and writing higher quality software in addition to dramatically more of it. Who has time for pablum?
How exactly are you using these tools that you have that experience?
"Find your path forward!" he shouted with glee, as he ran toward the cliff.
E.g how is the perf/$ vs Wildcat lake
So, on the mini the RAM upgrade runs at 25$ per GB on all tiers, the same as the Studio therefore the upgrade to 512 will probably cost 6400$.
The fully maxed out Apple Studio then will be 24699$. It's 17199$ if you don't upgrade the storage(1TB).
Nevertheless I itch to have one :)
EDIT: or buy AAPL. If I had bought Apple stock instead of buying a Mac LC II in 1992, then I would have about $2 million in Apple stock.
For general inference there’s no ROI that makes this work vs subscriptions.
25k for computer now, plus 9-10% sales tax, plus operating cost, plus time and cost for R&D tinkering with models, harnesses, and infra (assuming highly capable engineering talent that can get paid for your human inference) vs a HEAVILY subsidized subscription at 200 per month with free R&D has a pretty long ROI (15 years?)
At API costs, it’s like 6 months if you’re heavy on inference. For training, specialized models will have their own ROI that makes this worthwhile. Then debate renting capacity and the platform to choose
Unless you need privacy for your inference this instant, paying for credits can get 80 to 90 percent of people everything they need.
Of course if you do need that privacy, then forking the $25K over to Apple is a no brainer.
There are both cheaper and faster options out there.
Put together a similar build with a couple of rtx 6000 Ada cards and Apple's price tag suddenly looks pretty damn reasonable
"According to reports from Bloomberg, Apple will be skipping its M6 Pro, M6 Max, and M6 Ultra chips to accelerate development of the M7 chip. That means the only chip to be released from the M6 family will be the base M6.
The reason for this break with tradition: AI. Apple had been planning major neural-processing upgrades for the M7 family and ultimately decided those improvements were important enough to justify accelerating the next generation rather than completing the M6 lineup." https://9to5mac.com/2026/08/08/apple-m7-chip-heres-why-it-ma...
I'd skip M5 and M6 chips for LLM work and wait for a year for M7.
I believe that the CPUs are actually limited by ram bandwidth more than the neural engine right when it comes to LLM processing?
Maybe the M7 introduces something new to get around the current ram bandwidth problems on the non-Ultra chips.
Please say more? Is it because it is a one-time cost, unlike a recurring subscription of Claude/Codex?
Also, a lot of companies are looking at how to run capable models locally to cut some of their (massive) cloud AI bills. An easy answer is worth a lot to them.
For me personally, not quite that valuable yet, but I think it's getting there quickly. Deepseek V4 Flash massively increased the value of local AI to me, to the point where it's displaced most of my Claude Code usage, its upcoming vision enabled version should bump it further, and it's only going to get better from there.
It's a lot faster, but a lot of it is also feeling free to discuss things I wouldn't be comfortable sending to Claude, with the idea that that info is now theirs in perpetuity. I got my genome fully sequenced recently (it's cheap now!), and I get a battery of blood tests every year. Wouldn't do processing on any of that with Claude, but local AI? Totally great.
And if I was running a company with a large cloud AI bill, I'd probably buy a wheelbarrow full of these macs. Cheaper, but also a more solid/predictable base to build on.
AI based tools are very useful here - thinks like object removable or cleanup etc, not just AI generation.
For example Apple mentioned performance increases for https://learn.foundry.com/nuke/content/reference_guide/air_n...
If Apple didn't sold these things they wouldn't make them but, also the level of marketing that Apple is talking about for AI is basically the new group they need to capture because the ones I just listed are already buying Macs and or easily to motivate with the other obvious CPU / GPU performance upgrades for code compilation, faster memory and video transcoding.
to answer your question : looking at the aftermarket availability of Apple's prior best and brightest : practically no one buys them.
"people here buy them" , well, 'here' is one of the most affluent groups of people in the world.
They're available as movie and television set pieces (undoubtedly disappearing into the home of someone close to the staff post-production), and for administrative/boss types that can slip the cost into a ledger somewhere that few will ever see.
It has been a hobby of mine every few years to check out the apple site and see how big I can option a machine. My record was when I was in high school years ago and was able to option some pro studio-ish apple desktop thing to like 61,000 usd out the door.
For one thing, you can’t tell from a movie what the specs are. A $999 Mac Studio looks exactly the same as a $20,000 one.
For another, Apple updates the industrial design on their products so rarely, a 6-year-old Mac, iMac or MacBook also looks nearly indistinguishable from a brand-new one.
That may not be many people, but there certainly will be some people who want to do that, and are willing to pay big bucks to do so.
It’s when self hosting and local hosting was the norm, and why it’s also starting to come back.
There will be workloads that can never touch a public cloud, and for it solutions like this are an option.
I've often felt there is tremendous value locked up in underutilized old computers. It would be interesting to see Apple in 3 years offering compute as a service using lease returns (or more likely, partnering with someone else to operate it (perhaps exclusively in secondary markets like China or India, to address political demands for local siting or jobs). Apple is in the best position to work around or even gap-fix older software/hardware limitations in a controlled environment, and now they can do so without cannibalizing new hardware sales.
not getting on that bandwagon but wasn't that not the most demanding game as its a just a nonstop cutscene.
Which one is it you can run local models on? I suppose the NPU only.
I wasn't able to debug network errors (restartin my Mac worked), Metal was missing low level disassembly / debugging tools (there is some hard to use UI), but the worst thing was the inflexible windowing system.
Even getting all the window handles on all screens/desktops with their titles and programs is impossible.
I just decided that I move to Omarchy 4 (basically Hyperland + QuickShell) + NVIDIA GPU, and I already was able to customize it more than my Mac in years.
I will miss Apple's hardware for sure, but not MacOS and the missing hardware documentation
For example when using PyTorch I wanted to try to speed up my NN kernel by 2x by just using half precision and haven't noticed any speedup at all. Also I was missing the easy to use GNU tools that had to be mixed with Apple's tools.
I loved using Arc browser as well, and I'm missing it, but I guess I will do without it somehow (Chrome's vertical tabs are just not the same).
My main program missing from going back to Linux was ChatGPT Desktop, but now it's there.
I just checked out Hammerspoon, I'm happy for you that you wrote it, and looks great, but it has the same problem that I had: for security reasons Apple stopped allowing the window APIs to get all important information on other workspaces. You can only do it with Accessibility API. I was trying to fight with it but have up.
It’s a bit depressing because it means that if I ever feel forced to switch my daily driver, it won’t come without a dump truck load of friction, frustration, and lost productivity, which I’ve validated by using the various Linux desktops on secondary machines.
It's still not well integrated of course as those plugins are from different people, but I at least don't feel powerless as I know I can make any change easily.
I ordered an ASUS Zephyrus G16 with 5090 NVIDIA card + 1.9kg (quite an overkill, and I know that I will have to limit power output), but hasn't arrived yet.
But what's fun is that I love QML+QuickShell with its hot reloading, Hyprland with its Lua support.
With AI nowdays it's just so easy to do deep UI changes that wasn't possible a year ago.
In US its $4000 upgade so $25 for 1GB.
Also:
> 512GB memory option for M5 Ultra coming late October
If you want to comfortably afford this gen you had to trade options on memory stocks...
The problem I'm having now is that no models are targeting RAM of that size. Everything is either much smaller, targeting laptops, or much larger, targeting hardware well out of reach of enthusiasts.
Please, AI people, start making models targeting 128GB machines again. The last interesting one was Qwen 3.5 122B.
Should bench better than Opus 4.7.
Interesting times, to say the least!
You might expect the M5 Ultra to produce 50 t/s from Qwen 3.8 27B with a good context length.
I plan on maximizing my residual student benefits, and taking advantage of education pricing.
I edit 4K ProRes and H.265 footage, sometimes with multicam (up to 4 streams) and color adjustments, titles, etc. It's only after stacking 3-5 effects before things can stutter, really.
Or if you try doing something CPU-intense in the background _while_ running some heavy creative software. I just don't do that.
The prices are ridiculous though. I may just keep rolling with my Windows 10 setup.
That's wild!
Just amazing engineering push, the competition got the message and we benefit.
This may depend on the size of your library. I tried installing Jellyfin on a Synology NAS, which runs Plex just fine, and it ran so poorly it was basically unusable. It “worked”, but it was painful.
I can't believe that Apple still comes with this bullshit like 32 GBs is a lot. It's a lot for video memory - vRAM, but not RAM.
Also: "a staggering 1.2TB/s of unified memory bandwidth" -- yay, the GPU has reached the year 2020! (I'm a bit bitter that my M4 Max is near useless for local LLMs because of its low memory bandwidth.)
A m4 mac mini is better than al of these per dollar, msrp adjusted.
Hopefully by the end of the decade China figures out manufacturing at scale and fixes this.
Apple never needed to participate in the AI race to zero. Because they were already at the finish line years ago building their own chips that can run large >100B parameter AI models locally.
It's possible that they're working on their own LLM that's going to work very well on their chips, and possibly outperform anything out there when they do release it.
10 years ago 32GB ram laptops sounded too much. 8 was enough. These days even I would get that much ram since it’s soldered. 64GB is higher end.
In a few years we should see such high end hardware commonplace. Working with a local LLM to get work done is the ideal way to go which has mostly hardware limitation as of now that gets solved in due time.
Ten years ago I got 64gb of ram in my laptop, same as I have now. I bought both for business and personal use. System ram capacity hasn't changed much in 10 years.
It makes me curious how old you were 10 years ago.
We were definitely outliers that long ago. I put 64 GB in a MacBook Pro back in 2019, and that was (a) overkill for everything I ever ran on that machine, and (b) stupidly expensive by 2019 standards (albeit almost affordable by 2026 standards)
The iPhone 15 was almost entirely marketed based upon AI (I would say fraudulently so, advertising features they still haven't delivered), and a huge portion of the OS work was on local AI or AI integration.
And for that matter Apple has been dumping enormous sums into their own AI development. Their failure to have a lot to show for it doesn't void the fact that they tried really, really hard.
It's bizarre how often this "Apple sat on the sidelines and let the AI people fight...so smart!" narrative appears on HN. Apple hasn't gone down the path of spending hundreds of billions on nvidia GPU data centres, but they absolutely tried really hard to matter in AI.
Yep, Siri AI; they’re doing it in public.
But you get a generic computer and much more RAM.
And you lose a couple of organs.
It would take Apple one or two engineers to make Linux life much easier on macs. But Linux is outside their walled garden so it's ignored.
My Mac Mini is strictly a headless server for llama.cpp.
I use a Linux workstation.
If I were limited to use Mac hardware , I would install Linux in VMware Fusion and work from there.
My old 3090 is typically significantly faster (almost 2x token/s) than my M4 Max 128GB machine, as long as the model fits in the 24GB of VRAM.
In most situations it's a better idea to just buy tokens. But there are definitely cases when that's not an option. And then a machine like the M5 Ultra can allow you to do things locally for a fairly limited budget. And in a simpler package to manage than a machine with multiple GPUs.
* the iPhone * the iPad * apple watch * airpods * unified memory laptops and computers
Those are all products that either created a category or changed that industry.
I mean this is not nvidia based right? It's all custom? So we can use it under Asahi perhaps?
I want to get something for my company to run local models, wondering what would be a good option.
Linux runs very well in a VM on macOS. There are many good options for this, some free and open source (QEMU, UTM, Lima, Colima), some proprietary (VMware Fusion, Parallels).
But Linux in a VM doesn't get access to the real GPU, so model performance is limited. Those running on the CPU perform well, and those needing the GPU don't.
However, macOS on M-series macs is excellent for local models. (Maybe not as excellent as a box full of the best nVidia GPUs, but still excellent).
So if you're getting Apple hardware, like Linux, and want to run all of it locally, a fine setup for a machine to run local models, with agentic characteristics:
- macOS running one of the many local model runners. I used to use Ollama and Whisper, and now use llama.cpp instead of Ollama. Others use LM Studio, oMLX, etc. Provide HTTP endpoints to access the models.
- Linux in a VM for overall control and orchestration, with standard VM settings, and bridged networking so it appears as its own machine on your network. Also, in here provide a robust shared file server for shared state. Use this VM as your desktop and primary access to the machine, if you like Linux.
- Linux in a VM to launch ephemeral, volatile containers, with the containers using a memory-only tmpfs overlay on top of a read-only Linux filesystem in a VM disk image, with tools in this filesystem. Alternatively, a writable Linux filesystem in a VM disk image, with disk buffering set to use macOS host buffering and discard fsync requests. These settings optimise for container disk performance for data that's only ephemeral which will be deleted soon or on system shutdown. (You can combined both VMs, but need to use two VM disks to get equivalent behaviour, and be careful about VM disk configuration of the two disks.)
- Containers spawned within that second Linux VM can be spawned very quickly and run quickly, so are ideal for LLM agents that need a quick sandbox. These sandboxes generally run faster than a macOS sandbox, despite being on the same machine with VM overhead, because Linux is faster at some things. Teach the LLMs to store files and memories they want to keep in the shared file server.
I just want my butt ugly repairable beast machine to do the same trick. Why is my ram not unified? I have an iGPU in my server, but it can't access the 64 GB ram (I got last year for 150 euro) directly or something? It's on the CPU right? Why did only Apple go for this architecture? So many questions...
Can we please kill the xcode. It is worst pile of garbage I have to use just to develop ios app.
Is this a joke?
>Additionally, M5 Ultra features a massive amount of high-bandwidth unified memory, up to 512GB
Now we're talking. But at what cost?
256 memory gets you to like 11k. So like 15-20k.
"M6 also introduces a Dual 16-core Neural Engine, providing up to 2x the peak compute over previous generations to make on-device AI workflows run even faster" .
FWIW, Ollama, LM Studio and Lemonade (and oMLX) also wrap Apple's MLX framework.