New Mac mini, featuring M6 and M5 Pro
apple.com457 points by runako 15 hours ago
457 points by runako 15 hours ago
So happy with my base M4 purchase for $499 when it released. I even snagged a 1TB replacement drive for $199 a year ago. Feels like I got the last train out.
I know it's at least a little irrational, but I am indescribably angry with the pricing trajectory of personal computing. I'm not all that young and for pretty much all my life the computers you as a random person could buy have consistently become massively more performant and massively cheaper over time. Even when the exponential performance increases started to taper off, every few years you could buy or build a new computer that was either cheaper or faster than the one you already had.
That trend of the last several decades has been completely destroyed in the last year or so. Sure you can buy a new computer that might be slightly better than one a few years old, but it will be significantly more expensive to the point where an upgrade is almost certainly not worth it unless money is no object. Even replacing a computer from the pre-AI era would be virtually impossible at anywhere near what it cost you to buy it new unless it was verging on truly ancient by computer standards. Computers have always been a counterpoint to inflation everywhere else, but now that seems to be not only no longer true, but computer costs are inflating significantly worse than other goods.
I find myself increasingly in favor of the movement against data centers. Not because I think data centers are bad on their own, but because of the hording of computing resources they represent and the impact that has on people trying to buy any sort of computing device for their personal use. The PC I built 2 years ago is basically irreplaceable at anywhere near the price I paid for it. That's a truly unacceptable state of affairs in personal computing.
I don't think it's an accident that the market is on a trajectory toward personal computer (and particularly storage) rental being the only economically viable game. There are significant players who will profit directly. Indirectly it helps companies for whom end users owning general purpose computers and storage threaten their business models.
The industry has little incentive to fix this problem and as the market shrinks so does the incentive.
We lived thru the PC revolution. It looks over now to me.
I feel like we are basically reliving the COVID era supply chain crunch, that it's basically inevitable that in 18-36 months more capacity will come online and prices will crash back to earth.
Whether OEMs like Apple pass these savings along to consumers is tbd though.
Memory orders for the whole 2027 were already reserved. There is little indication that in 2028 situation will be any better unless something drastic happens like AI crash.
And then the opposite can happen like an unknown drone with 100 kg warhead striking memory production facility.
Yeah… “unknown”. As if we don’t know any countries seeking to end democracy in Taiwan…
“It costs a lot to buy a new computer better than the one I have” is another way of saying “computers aren’t going obsolete as fast” which sounds like a great thing.
It also costs more to buy the same computer you already have, which is not a great thing.
Isn’t this just inflation? We are in a highly inflationary environment. Everything is going up in nominal terms.
I think their point is that computers used to defy inflation. Across the decades, computers were getting cheaper in an absolute sense (i.e. not factoring in inflation) and astonishingly cheaper in terms of dollars per unit of performance.
Even when the cost of the cheapest computers bottomed out, it likely had more to do with lower prices not making sense. At a certain point, the cost of things that encapsulate the computer and I/O (like the enclosure and screen) end up overwhelming the cost of the actual compute.
no, inflation is the cost of goods rising because money is worth less.
This is capitalism, as demand increases, so does price, regardless of the cogs.
Inflation is a few percent a year roughly. Prices for hardware have skyrocketed. Ram has jumped anywhere from 2-4x in cost. NVMe drives are still going for 2-3x what they cost about 2 years ago. Just go look at one of the amazon price trackers to see how bad it is.
TSMC, plus flash and memory manufacturers with their own facilities, have dramatically cut production of almost anything that isn't in demand for AI, plus Macs were just about the only way to get more than 16GB of 'VRAM' without spending multiple thousands of dollars, so Mac Minis and Studios with more than about 24GB of ram have been under months-long backorders, which is unprecedented. Tim Apple made is name in supply chain logistics. Even he can't logistic his way out of this. Making things worse, the RAM that Macs use is really fast stuff, nearly the speed of high end desktop GPUs.
Supply for anything that isn't exactly what NVIDIA, Meta, OpenAI, Anthropic, and MS want has pretty much evaporated. And even that stuff has skyrocketed in cost because, essentially, the race has turned into who can hoard the most of TSMC's and NVIDIA's production capacity in a particular year.
The pathetic thing is that it isn't even actual purchases in some cases. In some cases, it's OpenAI or Anthropic merely "promising" they will buy X% of a memory manufacturer's capacity.
I can't wait for all of this to come crashing down.For all this hardware that is useless for anything else because of its density and power consumption - to absolutely crater in price and become e-waste.
Same thing for me! Got it with the education discount when it came out in 2024 and, although I've been pretty disappointed with macOS, it's crazy to see that the price increased by $300 in less than 2 years, with no sign of the price increases slowing down.
What will happen once people are priced out of buying computers? I find this worrying.
I don't really understand this dooming. All market forecasters, memory bulk purchasers and the memory producers themselves have stated multiple times, unequivocally that the shortage will get worse in 2027 and then as new fabs come online the shortage will taper off as 2028 progresses.
We've been through.. 4? 5? memory boom-bust-boom cycles already. Nothing is pointing to this one being any different, aside it probably becoming the mother of all busts since AI isn't adding much productivity and as businesses realize this, they'll stop paying top dollar for access meaning AI vendors cancel all their reserved fab capacity.
The fabs already exist at that point so it's less of a loss for the memory fabs to crank out memory chips and have the RAM OEMs sell them at bargain bin prices. No pseudo-cartel pricing possible either with CMXT becoming a major player. $50 for 32GB-64GB of DDR5 in 2029 here we come babyy
>I don't really understand this dooming.
no one remembers the last time things got better rather than worse. shitcoin-driven scarcity seamlessly transitioned into COVID-driven scarcity and now into hyperscaler-driven scarcity. the chipmakers' cabal decided to meet increased demand with four digit margins rather than expanding production.
Some of us do. In 2011 Thailand floods that destroyed major HDD manufacturing capacity and drives were really expensive.
It's Deja vu for NVMe and SSD drives this time as prices are way up there.
I don't remember any previous price increase this significant or sustained. Relevant to this thread, I don't ever remember major computer companies like Apple having to significantly jack up their prices because hardware costs had gone up so significantly and stayed there.
Aside: I used to love reading qz. Anyone remember when it was a chat/imessage style mobile app? The news was super short and quirky.
Then they sold out and most recently literally sold out to this software company from my neck of the woods. Every single article now feels like it's written by an LLM optimized to write like a crypto/AI bro.
I used to use it daily and now not at all.
> AI isn't adding much productivity
Would love to know where this comes from. I am not sure about that. Could you point to data/anecdata?
https://fortune.com/2026/08/22/executives-ai-productivity-la...
There's been a few studies. Most find productivity gain from 'AI' to be negligible.
Seems hard to believe, I mean I'm not an AI absolutist but as a programmer it allows me to automate simple things like writing small patches or config changes in a matter of minutes where before I would have invested hours, so hard to believe that it won't at least bump productivity up by 20-60 % overall, it's also tremendously better at finding relevant content and curating it than me so that's another time saver unrelated to programming but relevant for any type of knowledge work. Friends in research and medecine and law also tell me that AI has drastically accelerated their rate of processing and ingesting any type of information and helped them to automate aspects of their jobs. So not sure but negligible isn't the word I would use here.
> Seems hard to believe […]
This goes back before AI, with computers not being found to be improving productivity numbers: Solow's paradox.
* https://en.wikipedia.org/wiki/Productivity_paradox
* https://www.brookings.edu/articles/the-solow-productivity-pa...
> it allows me to automate simple things like writing small patches or config changes in a matter of minutes where before I would have invested hours
One of the problems here is that things like this only count as "increased productivity" in the economic sense if it results in more income at the end of the funnel. Are you actually making more product faster in all your new-found free time?
Very few orgs I've worked in have actually been limited by the speed of programmers writing code. Usually it's communication overhead and management processes eating up most of the bandwidth - and those tend to expand to absorb any additional bandwidth.
Yup. And in my experience, AI has made the communication and decision making overhead significantly worse thanks to 100x longer docs, code that noone understands and enormous pressure by executives to magically produce software that just works.
That's just because most people are bad at communication to start with.
Look at all the shit github README before llm. Most repos didn't bother because it was effort. those few that did, were because of autistic focus. Of those, very few were "great" user documentation.
now llms are here, the effort barrier is gone, and now we're inundated with shit README that provide no value to users nor respect their time.
Once you start applying SKILL.md that revolve around fixing HOW to communicate value to users in a way that respects their time, you'll see this story change in noticeable places.
I doubt it will change everywhere because it also requires that you give a shit about doing this in the first place, so those people Ed Zitron talks about... business idiots won't care, because they're ultimately nihlist.
> One of the problems here is that things like this only count as "increased productivity" in the economic sense if it results in more income at the end of the funnel. Are you actually making more product faster in all your new-found free time?
This strikes me as a problem with how we define productivity in this context. If I can do my job in 1 hour, when it took me 8 before, am I not more productive in a rate-based sense? Sure, I might only work 5 hours in a week instead of 40, so my net production is the same, but my rate of productivity was certainly much higher.
Wealth created is still the same, and that's what an economist means by "productive" - you produced X amount of wealth (usually measured in dollar amounts).
It's like "work" meaning one thing to a normal person, but something very specific and counterintuitive in Newtonian physics (if you carry a fifty-pound weight in a circle and stop at your starting point, you've done no work).
I mean I can deliver the same software in one month that I delivered in maybe 3 months before and for many things I don't need a dedicated frontend developer, designer or writer anymore. So for me it's the case!
Even if it was true… does anyone really believe that companies will stop using AI?
Ask any solo founder and they will have a very different story
https://www.economist.com/business/2026/07/27/america-has-be...
> What will happen once people are priced out of buying computers?
If it persists for years people will have time to spin up more fabs and meet demand
> What will happen once people are priced out of buying computers? I find this worrying.
What happens is that people start buying Chinese mini PCs, some of those even have spyware preinstalled along with another malware that is Windows 11
This is a great route. Just yank or wipe the windows disk, install Linux, and it’s way more secure and performant than any preload windows would be. It probably runs games better too.
Chinese mini PCs are also affected: they're shipping with less RAM, at higher prices, or both.
> What will happen once people are priced out of buying computers? I find this worrying.
I’m sure some kind capitalist will be happy to rent them a thin client linked to a computer in the cloud.
Apple is already leasing phones/computers..........
My take: this was done to pull consumers back into the cycle of upgrading to every new release. iPhones already do so much more than most people need. With the RAM price hikes, even though there was an influx of sales (according to a store rep) initially, I think Apple was expected a significant decline; people just choosing to stick with their current device to avoid paying the premium.
Leasing not only allows Apple to keep "selling" people new devices, it will also mostly like accelerate the second-hand sales and resource recycling arms of the business. Consumers will be more willing to upgrade sooner and stay in perpetual payments vs having to buy out the device at lease end - and Apple gets back newer devices where the recycled tech is arguably more valuable.
Yeah the era of the personal computer is over.
Companies will prioritize components for cloud computing and lease us thin clients...
I’m sure some kind capitalist will be happy to rent them a thin client linked to a computer in the cloud.
Maybe we'll see a return to the late 1990's, when companies would give you a free computer as long as you allowed it to constantly display an ad banner on the top of the screen.
I think this was only promised once or twice and any company relying on it quickly went bust.
There were at least seven companies offering this. I personally knew two people who had them from two different companies, and I think they were also available at Best Buy and Circuit City.
But yes, they all went out of business during when the dot-com bubble burst.
Capitalism will lead to:
- Lots of options competing on quality, price, user experience, user needs
- Every marketplace innovation that we talk about on this platform
So somebody might offer exactly that and if it's a good option, people will use it. If it's not, they'll use an alternative.
Government regulation will lead to:
- Giant entrenched companies that are difficult to disrupt thanks to insider connections
- Lack of competition, price increases, terrible customer service and pretty much everything that "anti-capitalists" today complain about in regards to capitalism.
So, government regulation will lead to all bad things we see today in capitalism? But capitalism won't. This is not very sound logic.
Theory holds so far.
Everything government involvement touches gets more expensive. See healthcare, college tuition, student loan debt as some easy examples.
Two of those are the same thing and both are much cheaper in countries other than the US where the government is much more involved
That’s odd because spend on all of those things per capita is consistently lower with better outcomes where they’re socialized.
> See healthcare
Medicare is cheaper that US private insurance:
* https://www.kff.org/medicare/how-much-more-than-medicare-do-...
Also:
> A single-payer universal health care system could cover every American, save more than 100,000 lives a year, and still cost $1 trillion less than the system it would replace, according to a new preprint study led by researchers at the Yale School of Public Health.
* https://ysph.yale.edu/news-article/universal-health-coverage...
*gestures at prices*
gestures at supply and demand
Yeah. It's not working to lower prices.
Right now, demand exceeds supply. As new fabs come online, supply will rise. Eventually, either through greater efficiency of the models and inference systems or through the collapse of AI companies, the demand will drop.
Capitalism under the control of a few billionaires who can influence the entire market with FUD leads to: current market conditions.
> Government regulation will lead to: > > - Giant entrenched companies that are difficult to disrupt thanks to insider connections
Oh, you mean like Apple, Google, and Microsoft?
> - Lack of competition, price increases, terrible customer service and pretty much everything that "anti-capitalists" today complain about in regards to capitalism.
Have you looked around? This is already happening.
Yes, absolutely.
And it leads to comments like the parent blaming "capitalism"
Capitalism doesn't exist without government regulation. Who's going to enforce the contracts? Who's going to break up the giant entrenched companies rent-squeezing every cent out of their natural monopolies, and then using their illicit gains to crush any would-be competitors as they ooze into other markets?
Literally everything listed for 'government regulation will lead to' can and has happened in unchecked capitalism. Insider connections rule everything right now in this current regime of chest-beating 'capitalists.'
Capitalism and socialism aren't real things outside of freshman year econ textbooks.
Current government already doesn't enforce contracts. Ex, prime video ads. Further there's no point in having contract terms if they can be changed unilaterally at any time.
Contracts are worth exactly as much as you're lawyered up and willing to get them enforced. There is absolutely still a functioning courts system to enforce contracts.
The one notable exception is if you're a group of workers and you negotiate a contract with your employer: the state can and will interfere to void it if it prevents the workers from being properly exploited.
Maybe they will have to use/repair/upgrade the ones they already own?
Problem with that is so few of them are repairable these days. Laptops just kind of fall apart over time.
Thinkpad and Framework are notable brands for easily replaced components. They should run forever if you just replace moving parts like fans and keyboards.
Macs, as nice as the hardware is, are hostile to repair.
I got an Apple refurbished m4 Mac mini with 32gb ram and 512gb ssd in September last year for $1300 Canadian.
The m6 with same specs are $2000.
Also m4 Mac mini seem to now be in stock on their refurbished store.
I was not in need of a mac mini then and I’m not in need now, but damn I hope my laptop will last a long time because FML the best price/performance minipc of the last few years got wrecked and the entire market is in shambles
My personal laptop is a 2018 ThinkPad Carbon X1 Gen 6. It's eight years old now. Still running Linux pretty well, but I really need to figure out what to do for a replacement sooner than later...
I got a 32gb M4 macbook air with the lower CPU core count, right before things got stupid. Can't even buy that skew now.
I wish I picked one up last fall when the stores were trying to clear them out. Now you can’t even buy them in person.
Wait, their website only shows preorders for M6 or M5 Pro, and they have nothing about the M4. So do they currently have zero Mac minis in stock?
Yes. I looked just two days ago to purchase a M4 Mac mini (I had no idea they would refresh it), and estimated delivery times were '5-6 weeks', with the highest RAM configs unavailable. Other retailers still have some in stock though.
I feel that way about my GPU for normal gaming use. Now I’m afraid of it dying on me.
I don't really care about gaming, but there's a PC I got for free a while back with a 2060ti that overheats if I don't peg the fans at 100% and keep the case open. Would be nice if that holds on.
I’m currently hoping that my Intel Arc B580 doesn’t die on me cause even a 5060 Ti 16 GB has reached like 700 EUR price over here. I don’t think the prices will get better anytime soon.
Same with my M1 Air for being on the move, even notebooks would be more expensive than they’re worth.
I’m milking my M1 Air, too. It’s still going strong, and I just can’t justify replacing it. Would love 64GB or more memory, though.
I got an essentially free upgrade from an M1 air to an M4 air and let me tell you...there is zero difference in day to day usage. You've got another 5+ years of headroom at least.
Yep, for cpu I’m sure that’s true. It isn’t often that I wish the M1 was faster. But I need more memory. 16 GB was fine five years ago, but it’s getting cramped for me now.
Consider underclocking/undervolt it a little, especially if you aren’t using the GPU to its full potential.
It will last longer and hopefully the GPU prices will recover in a few years.
I hope
I'm kind of lost on what to do with it. I have better workstation PCs, and I couldn't find a good way to turn it into a media center - it would be a great living room PC - but couldn't find a BT remote friendly solution last time I checked. Anyone has a solution for this ? I don't want to be using mouse/keyboard every time I sit on the couch - something that would let me stream something like jellyfin fullscreen with a remote ?
Maybe try a Steam Controller? I've used the OG one. I think the device itself registers as a keyboard. Its trackpad controls the mouse cursor and the triggers act as the various mouse clicks.
If you are running Steam in the background, one of the buttons triggers an HUD keyboard, where the left trackpad types keys on the left side of the keyboard and the right trackpad for the right side of the keyboard. My guess is the new controller works the same way.
Also, some media center software, such as Kodi, have gamepad support. So a gamepad will feel completely natural to control it without having to resort to using the mouse and HUD keyboard.
There are various "air mouse" variants that properly work with MacOS as well. I have one with some media/navigation keys on the front, and a tiny keyboard on the back - kodi I can navigate just with the navigation keys, and if I want to do a bit more (but not get out the proper keyboard) I just flip it around, and can do that as well.
> I'm kind of lost on what to do with it.
Personal assistant (e.g. openclaw) in the mac ecosystem. You need external models to power it ofc, but cool to have all the Apple integration and local network in your assistant. The claw can be addressed in multiple ways and can command your local hardware if you so wish.
I almost bought one for my kids back then and now seriously regret that I didn't.
My son is using my old ~2011 Mac mini with Linux Mint on it. Slow but works.
[flagged]
Downvotes are probably because it's obvious they are talking about the pricing, nothing to do with quality.
It's a bit of an emotional goodbye to the super cheap Mac Minis we got to enjoy for a while. I got some for family members, we had some serving as iOS/macOS build servers sitting in a closet at my last job, they were fantastic.
Not that they're a bad deal now, but at European prices of over €1000 for M6/16GB/256GB it's a psychological barrier that's been broken.
When they discontinued the M4 with 16GB a few months ago I had literally just ordered one off Amazon which they promptly cancelled. It would have been 600 after exchange rate from GBP.
I can't justify a more than 2x price increase for what I'm guessing are marginal improvements. I want to replace my home lab but as it stands my underpowered NAS will have to continue as it has the past few years
If you adjust for inflation, is it a big change from the super cheap Mac Minis of the past? Probably still somewhat more expensive, but not as much as the nominal difference would imply.
Mac Mini M4 launched less than 2 years ago at $600 (US, $500 with edu discount; as low as $400 on general discounts). Now same config with M6 is $900.
And except at the beginning, the Mac Minis of the past weren’t famous for being cheap, but rather for being disappointingly, unattractively expensive and, over time, apparently increasingly forgotten by Apple.
Mac minis from 2005-2024 launched with entry-level models priced from $499-$699, with the outliers being the 2018 model at $799:
https://everymac.com/global-mac-prices/all-msrp-mac-mini-pri...
Though $599 for the M4 in 2024 seems like a local optimum.
same spec, same chip. I tought Apple would at least bump the RAM
You thought they'd bump the base RAM in the middle of the biggest RAM crisis in history?
> but at European prices of over €1000 for M6/16GB/256GB it's a psychological barrier that's been broken.
We should be buying European solutions anyway. Different psychological barrier that was broken long before.
One of the things that Steve Jobs got very right was "and you can order right now". Announcements where you have to wait a month to order throw away all of the hype. Even allowing pre-orders and shipping later is a better idea and so silly that Apple has started walking away from that.
So many things I would impulse buy, but I have to wait a month and so I don't.
Don't give them ideas...
Wonder how the cybertruck sales would have gone if tesla had done that.
(250,000 pre-orders, vs now ~5000 sales/quarter)
I would venture a guess that... other... things happening related to Telsa over the course of the couple of years between the announcement and when you could actually buy one had a much bigger impact on sales.
Well, I think the cybertruck was announced at least a year or two before they shipped. A month wait is one thing, but giving people literal years to think about and find flaws with your product is a really bad move.
I am shocked every time I see one of those driving in SF. street parking must be awful!
I drove one as a loaner.
It is big. No dashboard. square steering wheel without turn signals or gear stalk. The rear-view mirror doesn't see out unless you lower the cover in the back. The side mirrors are strangely shaped and small. The front corners are a mysterious mostly unpredictable distance out the front window.
But then... It has cameras in all directions including the front bumper. There is a top-down sort of heat-map display when you park. All 4 wheels can be turned for a tighter turning circle. and it can self-park.
So... I would feel embarrassed to be seen in it, but it was definitely an interesting experience.
At least in the UK, that's exactly what they're doing right now. They're allowing pre-orders, with shipping on Sept 22.
There are some options, though, that you can’t even pre-order until October.
Which options? I’ve tried a few different configurations, including one with 64gb of ram, and all of them appear eligible for preorder.
What I am missing in all of those benchmarks, how does the M6 compares to the M5 Pro.
I don't care how the M6 compares to the M1, even when the M6 2.3x compared to M1 in Excel spreadsheet. That is no measurement even in accounting. The dead time is not the Excel.
I am to only one?
The big difference you're not looking at is the memory bandwidth. The M5 Pro has nearly twice the one of the M6:
170GB/s memory bandwidth Configurable up to 32GB unified memory
307GB/s memory bandwidth Configurable up to 64GB unified memory
That is also the factor you want to take into account for AI inference, of course.
Both are still very low for AI btw.
Yep, even 600GB/s is barely usable and not fun.
Well, “low” is relative. Not everyone can spend a small car’s worth on a Studio…
But everyone can load 10 bucks on a deepseek flash API and have almost unlimited usage at decent speed.
you're looking at the wrong device.
- Apple M5 Ultra chip
- 30-core CPU with 10 super cores and 20 performance cores
- 64-core GPU
- Neural Accelerators
- Hardware-accelerated ray tracing
- 32-core Neural Engine
- 1.2TB/s memory bandwidth
one point 2 terrabyte a second memory bandwidth....
512gb coming later this year
but at the cost of a small car
Last time the benchmark numbers from apple were stupid and they still are.
I just wait for practical real live numbers with real stacks i use.
And btw. they reduced the memory bandwdith on the M4 in comparision to its previous version by 40gb/sec or so, you need to wait to see.
It basically means that every meaningful comparison produced an embarrassing improvement multiplier
Upgraded my work machine from an M1 MBP to an M4 Mac Mini a couple of years ago. I do a lot of work in Excel with large sheets. The performance improvement was very noticeable and appreciated!
Benchmarks from the manufacturer are just marketing as well. Wait for independent 3rd party reviews.
https://www.apple.com/newsroom/2026/08/apple-introduces-m6-a...
Says 1.2x for multithreaded and doesn't specify (I presume smaller) for single-threaded. That's compared to M5 not M5 Pro, so M5 Pro is likely significantly faster.
> for always-on agentic computing
Don’t know, but this as main ad headline feels almost like a threat to my personal home computing needs
Don't worry, there are exactly zero Apple chips that are good for any sort of local inference right now. Unified ram is fast for ram, but dogshit slow compared to actual dedicated VRAM on graphics cards.
Your statement is not really true any more.
I have an old M1 Max with 32GB that runs inference models fine, especially those using MLX.
Dedicated graphics cards are faster but not to the point that it matters.
That’s a six year old machine.
My newer machine M5 max 128GB will far outperform your typical 32GB gaming card once the model exceeds memory.
To even get close to that on a dedicated graphics card you are paying upwards of $20K.
That said, models are getting smaller and faster which allows me to run multiple different models with ease.
[edit] Just checked the 5 models I use take up 57GB when all loaded at the same time.
The $8k machine will out preform the 1k gaming machine sure but it wont out preform anything in a similar price range.
> Dedicated graphics cards are faster but not to the point that it matters.
It definitely matters, if you're intending to run a Claude Code/OpenCode style agent workflow. Most of those harnesses start with 8-12k token contexts, which is a lot of prefill for a Mac but cheap for a CUDA GPU. The GPGPU compute on the fastest Macs is still trailing behind Nvidia's laptop GPUs; the highest-bandwidth Apple Silicon chip (now the M5 Ultra) has ~7x lower memory bandwidth than a single B100 card.
There's a good reason why Apple Silicon isn't to be found anywhere in the datacenter buildout. It's nonviable for training, and wastes electricity running real-world inference workloads.
Is it too much to give the harness a moment to load up the context when starting a new session? Is the brief pause truly a deal breaker?
That selling point is probably more about using a claude code subscription than running your own models. Because of apple's lockin there are certain things you can only do on apple devices and a mac mini is useful for that
The M-chips are insanely powerful. So powerful for my regular use that I'm still on M1-generated devices and don't really need to update. Perhaps in a few years when both local-running LLMs are good enough on regular consumer hardware there'll be a good reason to upgrade.
As a disclaimer though, I haven't even tried to really run local models on the M1 as I also have a 'gaming PC' that can run them fine.
I feel much the same. I feel like the next time I upgrade I could splurge a bit more and get a really bigger CPU and storage to be even more productive with.
However the main reason I would upgrade is to run LLMs locally and RAM prices are making this a no-go. One day if RAM comes down to earth I'll upgrade. If not I'll stick to this old machine for a long time.
yeah my m1 mac studio shows no signs of stopping and i can't justify another upgrade until doing so really unlocks something new at a reasonable price (like running local llms that are competitive with the cloud services at a decent speed, which presumably the mac studio M5 Ultras can do with qwen 3.8 but the 256GB Ultra config is more than $10k!)
> Perhaps in a few years when both local-running LLMs are good enough on regular consumer hardware there'll be a good reason to upgrade.
I mean, local LLMs already are really good if you own Nvidia Cards, not sure what you are waiting for. 2x3090 will blow any current Mac out of the water in tok/sec
I was talking about upgrading to another M-series device. I actually mentioned in the comment that you're replying to that I do run them locally on my desktop. And I run them off of an AMD card, 7900XTX.
I'm not sure I'd consider a gaming PC 'regular consumer' hardware especially with today's prices. But yeah, it's definitely doable.
After watching a video about how the filesystems are > 2x faster (and testing this myself in my mac mini vs a linux machine), I feel like Apple is leaving a lot on the table, and I hope that they get APFS to be as fast as the linux filesystems.
Because otherwise a lot of this hardware (including their blazing fast SSDs) is wasted. Cuz my M1 mac mini is awesome otherwise.
I took a peek into APFS a bit ago [1] because I (stupidly) yanked a usb SSD before it was unmounted and corrupted it. Reading the documentation, it seems to have many "cool" features which just aren't fully exposed/utilized in the OS, and there is a lack of API tooling.
For example: it is copy-on-write, and my corruption happened in the superblock checkpoint. There is no built-in command in Disk Utility to "mount" an older checkpoint (and accepting a small amount of dataloss). Also the "First Aid" command allegedly runs volume repair top-down when it is recommended to do it inside-out.
Claude was super helpful at walking me through the APFS CLI tool to do some more advanced stuff that’s not in the GUI.
It might not be the fastest, but at least it supports setting/keeping file creation time - unlike the clusterf*ck we still have on basically all Linux filesystems unfortunately
You're blaming the wrong thing.
It's the Linux kernel that deliberately does not provide user land a way to change file creation times.
Meanwhile XNU/macOS does.
Is APFS that much slower than ZFS or btrfs (more comparable Linux file systems) or just ext4? Btrfs is noticeably slower than ext4, but at least all the Mac and Linux file systems are way faster than file operations on Windows :)
Here's a very recent video benchmarking APFS vs ZFS, btrfs, and XFS+: https://www.youtube.com/watch?v=4wVNFaFDIn8
Just watched that after I got home from work. I was pleasantly surprised with how in-depth it was. Thanks for sharing it! It does seem like the worst issues he was seeing had to do with parallel IO performance, which is exacerbated by the agent development use case, so for sequential file creation/deletion, APFS is bad but not horrible.
My father was a plumber in the 80s and bought our family an Amiga 1000 for $1,200. Today thats about $3800 in inflation adjusted money. I just bought my son a top of the line pc for $2,000. I get that it used to be cheaper, but value is still relatively good for these machines.
Except we've gotten so much better at making computers since the Amiga, its a travesty that things have gotten so bad for consumers. You know the ones that bail out the bubbles
> Up to 2x faster gaming performance with ray tracing in Cyberpunk 2077: Ultimate Edition when compared to Mac mini with M4.
Interesting that Apple quotes performances when playing Cyberpunk 2077 ! Maybe they will want to care more about gaming ?
And while i'm at it, I was wondering a few weeks ago if I should try this game on my M3 pro. Anyone has played it on Apple and has feedback ?
Apple does this thing where they pick a very old game and then boast about how well it runs. It's so bizarre.
Not like Mac gaming is good anyway...they have "Game Porting Toolkit" but it's not free and open like Proton so nobody will use it unless Apple pushes them. Removing Rosetta doesn't help either.
In fairness, Cyberpunk 2077 might be 5 years old but it is the modern day "Can it run Crysis". It is absolutely still a hard game to run.
In fairness, Cyberpunk 2077 is playable through Wine translation on a $400 base model Steam Deck.
The game gets a bad rap for the last-gen console performance, but even 2020's iGPUs could handle it: https://youtu.be/oUm5GmJcBPA
Apple also just can't give up the idea of selling games through the AppStore either and you end up with situations like with AC: Shadows where it was ported, but you must buy the macport via the app store. Your copy on steam does not include the mac port, so if you wan to play on both your Windows/Linux desktop and Mac, you have to buy it twice. So stupid.
Cyberpunk isn't old to macOS, they just released the native version a few months ago.
> Game Porting Toolkit" but it's not free and open like Proton so nobody will use it unless Apple pushes them.
I don't know what this means - game porting toolkit is free
From what I've seen they do care quite a bit about gaming, but not to the point where they're willing to actually do much about it, so I wouldn't expect a proton style compatibility layer anytime soon unfortunately
You can get Proton through CrossOver at least. I haven't done much testing but I know some people use it to do some basic gaming.
I play quite a few games on my Mac Mini via Crossover, but rarely anything particularly graphically intensive. Though I do see occasional buzz about improvements in the game porting toolkit leading to significant perf improvements when they make their way into Crossover.
cp2077 sits permanently at 60fps and looks great on my m1 max 64gb. I used to play it on my 4090, but once I figured out how to do redmods on macos I quit using it and play it on my mac to save power (I'm 100% solar powered and barely generate 600w, 4090 consumes 500+). In apples infinite wisdom they don't support egpus, so my 4090 egpu will never work with my mac, which was a really disappointing thing to discover.
It could work but that would require NVIDIA to release an arm64 macOS driver. Safe to say this will never happen.
> Interesting that Apple quotes performances when playing Cyberpunk 2077 ! Maybe they will want to care more about gaming ?
They include game metrics in their marketing often, because they know their customers care about gaming. But no—for whatever reason, it never translates into Apple caring about gaming. I don't get it, but that's the way it has been for a long time.
With the “game mode” and “game porting toolkit” in recent history, I think there’s one person at Apple who cares about gaming and every couple of years he escapes from the basement where they keep him locked up and manages to sneak something into the release branch
They’ve also said when they pull Rosetta for intel compatibility, it will still be available for old games to use, which is a huge improvement from their usual stance of “that game studio that hasn’t existed since 2008 should dust off their code and update it to our new architecture.”
they've done a bunch to make macs a better gaming platform, including the very nice game porting toolkit (https://developer.apple.com/games/game-porting-toolkit/) which you can use yourself to run pc games on the mac. All macs since m1 have very nice gpus that can handle most games made today, and the newer ones can run the latest AAA ones with good detail. At this point it's more that game publishers/studios don't care about the mac
The "4x AI performance" smells suspicious to me, M5 Pro does have slightly higher memory bandwidth than the M4 Pro, but there are also the new Neural Accelerators in the GPU cores, I've always wondered what they actually do, since the SoC has AMX (Apple Matrix Coprocessor) since the original M1 chip era.
So is this performance gain probably due to the new Neural Accelerators?
Yes, the "neural accelerators" in the M5 GPU cores really do 4x pre-fill performance over M4. You can find benchmarks online since M5 has been out for months now. I assume the M6 has whatever the next-gen version of those is. These are not the "neural processors" that have been there since M1 (those still exist) but are additionl matrix math accelerators inside the actual GPU cores. Like "tensor cores" on Nvidia GPUs.
Thank you, found some benchmarks and they look really promising. To my understanding those Neural Accelerators are like AMX but for GPUs. With those accelerators the GPU performance on a M5 Max in LLM inferencing would totally be on par with a 5090, that's quite impressive!
Which inference benchmark are you looking at? The prefill speeds should be comparable on some LLMs, but the 5090 has much a higher theoretical max decode speed.
Do we need a 3rd top level discussion on this?
2.5Gbe as standard is nice.
About damn time, Gigabit ethernet started shipping on desktop Macs over 20 years ago.
Eh, for what? Most people use it only for 'internet' and 1G full duplex is generally still enough for that. Happy for 2.5G though :)
The people using Mac mini/mac studios are very likely media editors who have a NAS on the local network. 1gbit ethernet is slower than wifi at this point and incredibly outdated. 2.5gbit ethernet is quite cheap and common now.
Yeah, the Mac mini is cheap in cost but the buyer of that system is more demanding than someone who buys a MacBook Air or iMac despite the fact that the Mac mini is the cheapest Mac.
The Mac Studio is Apple’s realization that the best Mac Pro was actually the Mac mini with some slight modifications.
My cable internet is 2.5G. So it’s nice to get that full speed when I have a wired connection.
To be honest though, my usage does not benefit from the speed. The only benefit is to the reptile part of my brain that says “speed. Must have the fastest possible, and since I’m paying for it, even though maybe I should downgrade it, I want my speed.”
I’d love to take the extra speed if the networking equipment cost wasn’t so crazy.
Then, like you said, on the WAN side you’ve got the monthly internet bill.
1Gbps is the sweet spot, especially since 6Ghz WiFi runs at something like 600Mbps in the real world a reasonable distance away from the WiFi router in an ideal deployment. Saturate your wired, and you get a slight benefit over a 500Mbps plan for your wireless.
Sure, but I have 3 machines at home connected via 10G, and I do transfers between them often.
Double the price of the base m4 mac mini in February :(
(it was late in it's lifecycle and there were frequent sales)
I will buy a Mac mini, 64GB RAM + 2TB. That will replace my current MBP M1 Pro with 16GB (4 years old). For serious upgrade I will wait the M7 Line.
Considering doing exactly this, hopefully by the time M7 Ultra is out, the RAM Apocalypse will be over.
I don't know why some guy downvoted me. It's a great choice without expending 7K on some hardware than in less than two years will be replaced by the M7 (which is in the making).
I'm not a big Mac guy, but I have a couple of laptops. How do people feel about these? Use cases?
Which is more important for AI/LLM stuff, the number of GPU cores or the amount of unified memory? I can choose from a Mini with the M5 Pro 18CPU/30GPU and 64GB RAM or a Studio with a M5 Max 18CPU/40GPU and 48GB RAM for roughly the same price
They are both somewhat important. The compute speed impacts the tokens per second, and memory impacts the size of the model/capability.
IMO memory size is the most important since you can always wait longer for it to compute while a blazing fast but dumb model is much less useful.
Memory bandwidth is one of the most important factors. The Max machines typically have higher bandwidth than Pro (you should look up these exact models to verify though). Having said that, you could run bigger models with 64GB.
I went with a 64GB M2 Max a few years back even though it stretched my budget and I'm glad I did. I'm not sure what I'd pick in this case. Tough call.
They affect different things. You want more powerful models -> more memory. You want the models that run to be faster -> more gpu cores.
Related? "Apple to skip high-end M6 Mac chips in favor of AI-focused M7 line" june 2026 https://news.ycombinator.com/item?id=48676795 380 comments
Not really. It's Gurman, once again throwing darts at the wall to see what sticks like a desperate roach begging for subscribers to his bullshit newsletter of guesses and predictions he'll flip in a month the opposite direction for even more clicks and bait.
Wow. Is OpenClaw still alive? Would their users spend so many for the new Mac mini?
How many tokens per second would you even get on them?
Afaik most openclaw users aren’t running their llm locally - just using a mac mini as an always-on sandbox for agents.
A sandbox with access to iMessage. Can't have your agent texting with green boxes.
This is why my wife's iPhone 6 is still better in real life usage than the latest flagship Android phone. Playing high-end 3D game vs not screwing up the group chat.
Cool, like a $5 VPS?
A $5 VPS is easy to block. It's way harder to block a stock mac mini from a residential IP address
Tailscale the vps into your home network, and have it use that as an exit node.
But why a Mac mini over a $30 raspberry pi on the same residential connection?
Guess cause it'll still do local builds and stuff, and probably do a lot more of those if the agents are spamming builds and it's something like Swift or Rust that takes a lot to compile
OpenClaw was originally designed to run on a Mac, but using remote LLMs accessed via API. Running the LLM locally on beefier hardware came later.
I see that my old M1 Mac Mini with 16 GB memory and 1 TB drive has held value over time, selling for $600 to $700.
A new Mac Mini with that much ram and storage costs about double now.
I really want to buy one for coding/dev work, but it's just too expensive. Not sure if the M4 MacBook will get a price cut anytime soon.
You will not feel a difference in coding/dev work between M1 to M6 in a relevant manner.
Two years ago maybe, when I wrote code by hand.
Today development work is limited by compile times and test run times, particularly with UI tests in the iOS simulator.
Hardware on the development machine finally maters.
I work on a big web monolith on an M5 Pro 48GB.
I'm usually running about ~4 instances of it locally in separate Git worktrees, which (in our setup) means I'm also running 4 copies of the whole stack: web, frontend, Redis, Postgres, etc. These are rebuilt and torn down multiple times per day. Lot of frontend assets to compile.
Additionally, like many people in corporate environments, we do of course also pay a performance tax thanks to the security software we need to run.
So perf does matter for us.
We could, of course, alternatively... just run cheap shitboxes locally and have all our dev environments in the cloud. Certainly appealing in many ways from a cost perspective and you don't even need to give up DX really.
However, in the next ~5 years, as "actually useful" local LLM inference starts to become genuinely useful (for software engineering, this seems to roughly mean "the ability to run 28B+ models at usable speeds") then having local compute horsepower might start to look more attractive again. Even if it's not an absolute winner on cost/utility alone, the data privacy and cost certainty have value to many.
I have an M1 and M4 Max at office. I feel the sluggishness of M1 now, after years it finally started to show its age.
Maybe try reinstalling macos? I have an M1 pro personally and an M4 pro for work and I don't feel any difference between them, but I reinstalled a fresh copy of macos recently.
I feel like a big part of macos slowing down is installing too much crap over the years that sits in the background.
Have been considering a used M1 but I'm concerned that I'm going to get flash storage that's been thrashed to within an inch of it's life by someone running it as a poorly configured inference host.
I'm praying Framework becomes more affordable after the ram price inflation is over. Because I never want to buy another computer with non replaceable storage again.
There are sooo sooo many students and normal people using these devices or corporate laptops never seen inferencing or anything like this, its so hard to trash the storage of them, i wouldn't worry about it.
Where's a reliable place to buy a used Mac? I'm afraid of ending up with a broken unit.
Last week I purchased a MacBook Pro M1 Max from www.macsales.com, which is owned by Other World Computer, a maker of Mac peripherals. My computer comes with a 1-year warranty. It was my first time purchasing from them.
Yeah, I know. The problem is that used prices right now are almost the same as what they used to cost brand new. Prices have gone up way too much.
I also can feel a speed/sluggishness difference between my M1 Max MacBook Pro and my M4 Pro (14 core).
Honest question: Which one feels faster?
Obviously the M4 Pro feels faster.
It’s not obvious to me. It’s somewhat surprising, even, given the specs I just looked up:
M1 Max: 400 GB/s memory bandwidth, ~20.8 FP16 Tflops
M4 Pro: 273 GB/s memory bandwidth ~17.2 FP16 Tflops
True but memory bandwidth is not the entire story. Geekbench (while not a end-all benchmark) shows:
Single-Core Multi-Core
M1 Max, 10-core 2,204 14,004
M4 Pro, 14-core 3,311 24,711
Makes sense. I am still getting a feel for the different levels of these processors.
Is the "Water Project" intended as a sly dig or some kind of nod to AI water use concerns?
So I guess M5 Pro is better, right?
I guess so. It's slightly cheaper to buy 2x 32GB M6 minis than opt for 64GB sku (which is only available with the M5 Pro chip)...
price hike makes the deal not so sweet, but still a bang on device
Not a surprise with John Ternus taking the reigns as CEO in a week, who was previously SVP of hardware engineering.
I can't hate a direction where Apple becomes more about building great computers rather than trying to force more and more subscriptions. I do wish they would fix many of the long-standing OS and native app problems.
I do worry that at this point it's no longer just Apple's problem to solve. They need to figure out a way to take the rest of the industry with it. Maybe if their standards are high enough everyone else will look bad by comparison.
My daily driver personal machines are incredible pieces of kit, but everything is always getting worse because majority of software is some SaaS living in the cloud, or even if on-desktop is constantly updating slop.
Why does it take 10 seconds for my streaming music app to open when my machine is 10-20 cores, 32GB+ RAM, high bandwidth NVMe storage, and I am connected to gigabit fiber internet? The thing is autoplaying videos, and has about 10 different panes clearly owned by different product managers that hate each other. Half the time its pushing me some pop-up dialog I don't care about.
Winamp in 1998 took about 2 seconds to load.
I agree with you, but I also think sadly most consumers don't really care about this. The reason the internet is in the state it's in is because most people tolerate it.
Most people simply don't care about using well-made software so long as it gets the job done because they don't even look at other options. I suppose this is how carpenters feel when they see ubiquitous IKEA furniture.
I don't think polishing macos more will win more users, Windows is a flaming garbage heap where virtually every built in program is broken and people keep using it. Making macbooks cheaper and more accessible will do far more.
I’m still not convinced by the value of local models. It feels like ray-tracing where we are driving up the compute cost for a high cost and fringe benefit.
It's not worth it yet when datacenter compute is being subsidized and consumer hardware is massively overpriced, but the tipping point will come where it makes far more sense.
Particularly with Chinese companies pushing to get more out of smaller models. You can run really capable models on the top end macs right now, it just costs way more than buying credits.
I’m not really sure what the end goal is with Apple investing so heavily in local AI hardware.
Feels like they’re doing it for the sake of getting with the times, but why would the average Mac User care? I wonder if this an answer the OpenClaw craze.
Running useful models locally with privacy would be a great selling point. We're not quite there, but the next five years should be interesting.
glad that i purchased mac studio this January
Very well timed for John Ternus's first quarter.
I'm so sick of "up to 2x faster" ...THAN WHAT? ON WHAT WORKLOAD? How can I reproduce these claims?
I really miss when vendors would run the SPEC benchmark suite and performance was quantitative, and not qualitative.
… from the fine print:
“Results are compared to previous-generation Mac mini systems with Apple M4, 10-core CPU, 10-core GPU, 32GB of unified memory, and 2TB SSD.”
Running what workload? How can I reproduce it?
There is more detail in the footnotes found here: https://www.apple.com/mac-mini/
Upgrading to 48GB memory is $600…. Gut punch. I have a M4 Pro (top cpu skew 14 cores). Any idea the performance gain to M5 Pro? I feel like they holding off releasing M6 Pro because a new studio is dropping soon as well.
hmm waited for this, but the price... puuh
The "up to" and "Yx" compared to three generations ago specs from Apple. Never change.
I think there's a valid argument for this pattern when you know most people are on 3-ish year upgrade cycles. Also, if you compare against the most recent generation, you risk creating buyer's remorse for recent purchases.
I'd imagine most are on longer than 3 years for computers. Most (non developers) I know hang onto their laptops and computers until they die or get spec constrained, and that ends up being around ~6 or 7 years, so in this case the comparison to the M1/M1 Max makes sense, as for the next couple years that's going to be the cohort that's looking to upgrade.
The people who read Apple press posts would skew towards upgrading more often. Ultimately there is no number that would satisfy everyone and you are reading a marketing piece from Apple.
Wait for the product to be publicly available if you want in depth testing.
Sigh. I had the m4 in my cart yesterday and now it is gone.
~75% higher price for some extra GPU power and small CPU boost compared to M4? Eh. It's pretty far from M4's value, which was a no-brainer recommendation for people who wanted a tiny box sitting somewhere on their desk.
So the price just went up by another $100 am I reading that right?
Yup. Yesterday it was $799 for the base Mac Mini M4, a $200 increase from the original MSRP before June’s price hikes.
I paid $499 for my M4 back in April with the academic discount since I predicted the next generation base model would cost more. Sadly my prediction was correct: $599 to $899 is steep. I understand why; the prices of DIMMs, GPUs, and SSDs have also skyrocketed. But $899 for a base Mac Mini is painful, even if the alternatives on the current market are also painfully expensive.
Era of cheap compute is over I'm afraid
No, it's not. We do have to live through this long and harsh winter, though.
I really want a mini for running local models, but there's no way I'm buying one for the next few years. I'm planning on waiting for models to get better and more powerful, as well as for hardware prices to come down.
Over for now, or at least a little while until production ramps up to our new reality (~5 years minimum is likely).
Pretty much. Everyone seems to think prices will come back down, but I don't agree.
No 8GB option? Oh the horror.
Now if only it would come with Linux instead of MacOS!
Apple's software holds the hardware back, and it's ugly (especially liquid glass!)
Dear mods, please could these announcements be merged, if not done away with? They aren't really following the guidelines, they aren't 'curiosity', these are little more than marketing announcements.