New 14‑inch MacBook Pro with M6
apple.com
apple.com
I think the M1+ MacBook Airs will be regarded eventually like the original Macintosh originally the iPod, or the introduction of the original Air and the creation of the product class that came to be known as “ultrabooks”. It really changed everything about laptops to have a fantastic keyboard, speakers, screen, trackpad, hinge, and 12+ hour battery life… for $1k.
But with all of the AI-discovered exploits I think it'll become more important to keep everything up to date
My 2013 Core i7 took almost eight years to get to where the previous ones were at five, and if it weren’t for local LLMs pushing me toward an upgrade in the next year or two, I have to imagine my M1 Pro could be my daily driver well into the next decade.
Apple Silicon is certainly incredible (I also have a base-model M4mini, but it severely lacks RAM/ports so I don't really use it much/eWaste); my only Silicon-purchase-regret was buying a beautiful new 2024 MacBookAir (15") – heavily discounted $850@CostCo – but with only 8GB of RAM (it is functionally worthless, other than for watching it destroy your SSD with page-outs as you casually stream YouTubies...). In 2024, only 8GB of RAM (soldered onto the board because fuck you)!
This will be the “640kb ought to be enough for anybody” Bill Gates quote of our era.
I'll go and look at this new box, drool a little, and realize that I'd get more personal utility from donating $3k (or whatever the trade in delta is) to a charity than upgrading my computer.
I, for one, prefer Apple products because they sweat the small stuff more than the other options, and when they have put in effort to make things worse it's kind of infuriating.
I do a bit of desktop Linux work too, and it's OK, but hard to complain when it's all open source volunteer work! Windows is so terrible that I wouldn't even know where to start. What's the point? There's nobody at Microsoft that even gives a damn.
Apple was actually last to the party on that. Linux had virtual desktops for years at that point, and Windows had them as well through Microsoft PowerToys. I’m pretty sure I was running it on Windows 2000.
In the end you’ll spend a significant amount of your time trying to “get back” to where you just where (that alt-tab, a possible savior for this in all other OSs, doesn’t work at all on Mac just adds to the UX mess)
In the end though I just feel like it's extra cognitive load to have to think about two different shortcuts to do what is in many ways the same thing.
Say, I've got two browser windows open and a code editor open. I'm wanting to flip through the browser windows to check on things while I work in the code editor. In reality I'm just wanting to juggle the three windows, but depending on if I'm wanting to go from browser to browser it's a different shortcut than if I'm wanting to go from editor to browser. And there's a good chance I'll then need to swap from editor to browser, then issue a different shortcut to go from that browser to the other. When I could have just hit the same shortcut twice.
I've just never really encountered a situation where I've felt like it added something, but I'm constantly having cases where its more annoying.
The 40-year-deep-history Mac answer to workflows like that with a small working set with frequent cross-app-one-to-other-switching IMO is mouse-based navigation + spatial partially-overlapping windows, but this seems to be an extremely foreign concept to Windows/*nix users that grew up on full-screen-maximize or tiling. So I don't think is particularly likely to be adopted by anyone who doesn't already have it burned into their brains at this point. :) There's a lot of these old Mac bones still baked into the OS (e.g. if you have a folder on your desktop, and double click it, and move it, then close it, it'll open to the location you moved it to next time. And that memory is separate from that for another folder on your desktop. Though some of the follow-on behaviors have changed (opening folders in already-open windows used to behave similarly, opening new windows by default).
The 20-year-history version is my own personal lord and savior: Exposé hot corners.
Spaces and no-mouse navigation are still in many ways comparative afterthoughts to mouse-based/spatial workflows. Though as someone with 30+ year of Mac muscle memories in my head, the switching-windows-vs-switching-app concept and awareness of "where" I am relative to it is pretty embedded.
The whole window arranging so you can click on them doesn't help you when you have multiple desktops, close some terminal, and jump to some other terminal three desktops away. Like, arranging windows so you see some part of it always, just so you can click on it to switch back and forth between windows feels like a horrible workaround for broken alt-tab behavior. But there's additional levels of weirdnesses that just make using Apple frustrating and scary, like it's constantly throwing random curve balls. It just reeks disorganization. Which is really unfortunate - 10 years ago or so (when witcher task switcher still worked) I really enjoyed working mac. Back then I was mostly coding in the same windows I guess. Nowadays it's business work, and it's just a pain.
I get that that behaviour shouldn’t just be changed but I would appreciate it if there were another key combination that would allow cycling through all windows of an app.
Expose + Cmd-tab/Cmd-backtick being separated == a gap that Linux/Windows have somehow never managed to close in 20+ years now.
The quantity of requests for how to disable Liquid Glass on Apple Discussions is staggering. [2]
[1]https://www.macrumors.com/2026/06/09/macos-golden-gate-liqui...
This eventually becomes awkwardly precise complaints about information density and general change; but the underlying issue is I want to go fast and the system says no
On the Mac side, it’s a toss up between the platinum interface of Mac OS 8/9 and Mac OS X Tiger. With that said, I have a work-issued MacBook Pro running Golden Gate as my daily driver, and Liquid Glass is alright. It’s not what I’d prefer if I had a choice, but it’s not a dramatic departure from previous versions of macOS. A Cheetah or Puma Mac OS X user should be able to navigate Golden Gate with ease, mostly.
I've been on the lookout for a good condition G3 beige desktop, that was my first ever computer that was "mine" and I really want one just for the sentimental value.
Like GP, I despise the new design. I'll say the old design was pretty peak design. Windows looked like windows. Buttons like buttons. With accessibility tweaks (for transparency and borders) I could make it very, very usable.
This new thing is a crime against UI. Every icon now sits in a circle, and takes up % more room as a result. Because … it's pretty? It isn't pretty.
I upgraded my laptop thinking I was going to develop an app and that I'd need to be on the latest design. I ended up developing a TUI so I needn't have bothered. Every time I see my partner's laptop, still on Sequoia, I shed a little tear.
Depends on how accurate you need to be. How fast/confident you can be at hitting with a finger can easily drop off a cliff.
This is likely true for casual users, but for veterans this is a skill issue. My touchpad accuracy/speed more than matches most folks touchscreen accuracy/speed.
Skilled users can make the reach and the motion one. You don't need to stabilize your palm on the rest to use the trackpad. Same way I don't need to use F and J for the keyboard (and kinda wish I could remove those annoying bumps)
I am moving my finger on my touchpad to move the pointer to more or less match what I am reading. I am always close to a Reply button or a vote one. There is no point reaching for the screen with my hand and touching those links. Furthermore they are way too small to be pressed with a finger. They are probably too small for a pointer too, but that's another issue.
Hoever, when I open an empty tab, type new in the URL bar and FF completes it to the URL of this site, and I press return, I usually end up with a Login link in the top right. If I reload the page (ctrl-r) I am logged in. For some reason either FF is not sending the session cookie or HN is not doing what it should on a cold session. This is also another issue (sorry) but let's say that I want to login. My pointer is far away from the login link. My fingers are not on the touchpad. Is it faster to move my right hand towards myself, touch the touchpad, slide the pointer to the link or moving that very same index finger to the link and touch it? My laptop does not have a touchscreen but when I used a tablet with a mouse and a keyboard I reached for the screen many times. Of course a mouse is less efficient than a touchpad, except for games, but I also miss the touchscreen sometimes.
And I would use if to press the reply button below this line I am typing. Let's see where the pointer ended up while I was typing... At the top right corner. Moving it to Reply and clicking is slower than touching the screen.
What are you doing to them?
I still have MBPs from 2015 and 2018 that have not 'died'.
2012 MacBook Pro
2015 MacBook Pro
2019 MacBook Air
2020 M1 MacBook Pro
I almost wish one of them would konk out so I can get an M series, but then I wouldn’t be on a 5k iMac anymore, which would suck. Love these machines.
I get that this is hn and we're paranoid about everything, but come on, what CVE is a threat to joe sixpack here where he should drop $2K on a new machine to protect against it?
Dropbox would fire up to do massive walks of the filesystem tree and the fans would take off. I think the thermal interface material on the GPU got cooked at some point. It would be a relatively easy fix, for me a couple of hours, but I chose to fix it with an M1 MacBook Air.
The battery doesn't hold a charge anymore, but as long as it's plugged in, it keeps chugging along.
You gotta be kidding me. The old M1 Air had the "bendgate" where several people, me included, saw their screen die overnight for no reason. Tens and tens of pages of people complaining about it only to be with "you're holding it wrong" answers. Insanity.
I think Apple fixed that mess with the gen after that one.
(We replaced it with a Framework... with removable storage.)
I'm assuming that all of my local work would be lost. But I'd still have the sixteen million duplicate photos.
iCloud's priorities are not a great fit for preserving my source code repos. Fortunately those aren't very big.
But when my son's Mac dies, his huge Logic Pro projects are toast. And he is a few thousand miles away now.
However, OneDrive doesn't seem to even know how to handle source files (it doesn't even attempt to honor the gitignore), so rebuilding a project in folders covered by OneDrive results in flurries of temporary files being uploaded/updated to the cloud, hogging the bandwidth, with a bonus of locking files, crashing builds, or even reverting changes silently.
One would think backup tools, even if primarily for normal users are still literally made by programmers, at least consider the use case of programmers instead of just ignoring them entirely.
file sync is backup against the things most people have to deal with (hardware breaking, theft, house fire, residential lightning strike)
I would say accidentally deleting a file is more common than a house fire, so NO, file-sync is not a backup nor is RAID (block-sync) or clustering (data/state-sync).
A backup is/should be a non-modifiable snapshot of data/files maybe even a state.
I give you an example:
You have a password-safe (as a file). Your computer blows up, but you are safe, because your password-safe is synced on Dropbox....but oh my god what was the password to Dropbox??? Your restore is gone because you forgot to "backup" your password-safe.
However you are not a Loser, being a Admin is a ~lost art and a really good paid one (again).... ;)
It goes to your trash can/recycle bin for a while. That's literally what it's for: to guard against accidental deletes. Very few people accidentally delete files from their computer's drive anymore. The only time people do this in 2026 is accidentally formatting an SD card that had their camera's pictures on it or something.
I also do a bunch of video work with Da Vinci Resolve, and those project files are in the same folder. I don't keep the clips there, those are on an external disc, but that's just because I haven't gone from the two terabyte iCloud plan to the six terabyte plan.
There is iCloud Drive sync, but that only syncs your Documents and Desktop folders, but any other files are not synced.
Really this should be devices full stop. Also, "backed up" should include no fewer than two backups on different media and in different physical locations e.g. Time Machine to an external SSD _plus_ continuous cloud backup.
2/ This is like complaining that climate activists were harping too much 30 years ago, and that's why nobody listened
this is a serious issue though.
if you harp too much about something, it becomes background noise. people get desensitized.
if you harp about too many things, it fatigues people as well (ever heard of "alert fatigue"?).
so yeah, harping too much, too loudly about too many things is a serious and the society should really start looking into this
(edit: i am aware of the irony of harping about harping)
Or is the only way to do that successfully, doing it subconsciously/covertly like the state actors are swaying public opinion? :-/
No pain no gain also goes the other way, many times people don't learn until they get slapped in the face.
That'll be fun when half the world is migrating due to extreme climate events.
2. You mean the climate activists who used to risk their lives back in the 80s on open oceans instead of gluing themselves to streets? Yeah, the times have changed.
I suppose I didn’t specify exactly what I would recommend to someone who thinks it’s too hard, so here is that:
1. Install Backblaze. This is basically fire and forget. 2. (1) is enough for most people, but if interested (and on a Mac, which is what I’m familiar with), plug in an external SSD once a week or so for Time Machine. Setup involves clicking on the drive in the setup wizard and specifying an encryption password.
I don’t know about “ultra” robust, but this gives you protection from accidental deletions, natural disasters happening to your computer, hardware failures, and corruption from software bugs. The ongoing maintenance is plugging in a hard drive once a week.
If that is too much effort, just (1) provides natural disaster and hardware failure protection. It helps somewhat with the others but Backblaze only saves versions for 60 days and that’s a tight window to notice subtle corruption of e.g. your photos.
how many good things have 'average people' (and the avaricious wanterpreneurs chasing their dollars) ruined? Nearly everything!
> We replaced it with a Framework... with removable storage.
Cool. How are you backing it up?
How would you restore the backup if your secure hardware device is lost/stolen/damaged?
No point in encrypting those family photos burned to disc.
My macbook updates my documents etc to my NAS, and the NAS syncs to crucial folders to the cloud (not everything for cost reasons)
You should always assume your device can be stolen, you can spill fluids on it (or a guest), a cat can pee on it, an electric shock may kill it... No matter how great care you take of it, shit happens, data needs to be safe.
The SSD dying standalone vs the system dying doesn't change at all for data security. It's not a different risk factor
I wouldn't choose to buy a car I use every day that's 100x less reliable than a normal car.
If your looking for arguments against soldered storage the facts ssds are wear items and will eventually need replacement is a much better one.
If given a choice, I prefer a system where a HDD/RAM/GPU/CPU/PSU/Screen/Port/Battery/Keyboard failure is barely if at all disruptive to one where all of these effectively brick the system.
The point is it shouldn’t matter if the cpu or ssd dies because you have backed up everything already and are prepared for anything to fail or happen to the device rather then “oh the ssd is removable so in these select few situations I can recover my data”
Obviously if you used the same encryption scheme on a removable drive you'd hit the same issue - but most people don't do that, because storing the encryption keys in the Secure Enclave makes your drive effectively no longer removable.
Apple with their SoC really changed a lot of people's minds about hardware.
It's sad that M5 (and M6) are so much faster at prefilling and other LLM-related operations, but I'm happy with the computer I got and will keep it at least 3-4 generations.
- a low end Macbook Pro or Air
- Mac Studio, mid-to-high specs
My M1 is still fast and enjoyable, but I regret to inform you that the M5 actually does feel faster in day to day use. I thought for sure I wouldn’t notice, but the difference is there.
When I do large builds the difference is also bigger than I was expecting.
Also why would you have to spend $6000? You can get a refurbished 16-inch M5 Pro with 48 Gb RAM and 1 Tb SSD from Apple for $3000.
Mine is running great and it has 24/7 use.
https://github.com/nace-ai/kapitsa
this machine just keeps getting better and better in the inference world. the memory bandwidth is too good, newer macbook pros aren't that much faster in memory bandwidth, many are slower. not enough to upgrade yet as they're mostly horizontal upgrades, the improvements in other areas are adding up like wifi bandwidth and protocol, NVMe speeds, webcam
if they do the M7 correctly: which is just everything their current models do plus way faster memory bandwidth and much greater memory possible 192GB+, then it will be a really great and timely upgrade
M1 Max 400GB/s
M4 Pro 273GB/s
M5 Pro 307GB/s
I'll check it that inference engine! Thanks!
Edit: it doesn't give numbers for unsloth/Qwen-3.6-35B-A3B 4bit, but I'm using mlx-serve right now and getting petty good numbers (and it took a while to tweak everything properly!)
it has unsloth on the list of 3.6 35b variants only on the 2-bit unsloth UD-IQ2_XXS GGUF variant
I'd say let it down 4-bit Qwen 3.6 35b again and see what how it performs in both benchmarks and your preferred conversational and agentic harness
it's Kflash verifier is competing against MTP and Dflash2, thats where the additional novelty is
Has the webcam changed or improved in the last couple years? It seems to still have the same webcam specs as the M4?
Having said that, I did plan on sticking it out 7 years. Ended up bailing on an M5 Pro!
I like to factory reset my devices every couple of years, it really gives them a new lease on life. My M1 Pro feels as smooth and fast as the day I got it.
I replied that it would make no sense unless I was going to start running local inference or training my own models locally...
2 years later, and this MacBook Pro M4 Max w/48GB is the best $4k I've ever spent!
The engineering that went into the M1 was amazing, I truly think that now, an M1 Max was the biggest bargain ever in my entire computing life. £3,299.00. In contrast, the 64GB i9 I bought only two years before was £4,149.00.
The M1 Max MacBook with 64 gigabytes is still a great machine and fairly affordable on the second-hand market.
(Returning an i9 to wait a few more years on the M1, any M1, was absolutely worth it)
I have a 16" m4 max from work (36gb) and is an incredible machine (build quality, incredible display, overall feel). I also have a private macbook neo which should be similar to M1 in terms of performance but it's limited to 8gb of memory.
The neo/M1 performances are sufficient for my needs (for private computing) and chances are the M1 pro actually performs better than my neo (if anything for the more abundant memory).
I might to back to the listing and try to haggle the price even lower, considering a new generation of macbook pros has just been announced.
I do wish they had removable storage though, I originally got that Macbook as it was the cheapest way I could try out MacOS and M1, once I decided I liked it, I was stuck with 256gb or spending all that money (and then some) again. This time, I went with 1tb.
Still a great machine, shame those keycaps wear so badly (developing shiny greasy-looking patches)
authorized repair shop says 4 days to fix it and it's my only laptop, so i go without :(
(Obviously there are other reasons to solder like performance)
Until then, then M1 Pro will be fine for my usage, which is already more advanced than most.
My ideal scenario:
- 2027 the AI bubble bursts
- 2028 new memory factories reach the production stage and fail to meet demand, so we have a glut of memory and rock-bottom prices.
My dev machine is an i5 of some description about 6 years old, that's about all I could tell you off the top of my head because I've never worried about needing slightly faster compiles. (Ignoring running local LLMs of course. Currently they can't compete with Opus 5.5)
In my experience with this, most of the work happens on a HPC cluster. Your actual machine just needs to work as a thin client.
That said I don't do any local AI or 8K video editing type stuff.
On upgrade I could probably live on a moderate/high spec Mac mini since the only reason I went Studio originally was the resolution/number of displays it could support.
There are no features or performance characteristics of any of the subsequent M[2-6] Macbook Air or Pro models that I care about. I'm not really sold on running models on the hardware I have right in front of me. So long as I can rent an RTX PRO 4000 Blackwell at about half a buck per hour through a European provider, I'm fine using that with Qwen3.8-27B UD-Q4_K_M for the few things I consider more privacy-sensitive.
I've been daily driving it for months with no blockers. You can tell Linux uses the battery less efficiently than MacOS though so that's good to know.
So, 97% stable then? (Downvote me but it's true.)
That said, I do have hopes that AI will basically solve this. It's extremely good at diagnosing low level bugs, reverse engineering drivers and so on. A silver lining of AI taking our jobs at least!
It is perhaps also possible that others have different experiences with both Windows/Mac and Linux than you do. It's common for people to not notice what's broken, or what they have to workaround, with the software they use every day. People also have different workflows and priorities that determine whether their use of a tool is "rock solid" or not. (Something I certainly wouldn't say about Windows—or Mac!)
Windows is rock solid?
Meanwhile my Windows install can't see either of the wifi devices, even when I try to install drivers from the manufacturer. CachyOS autodetects everything and it just works.
So, not pretending, I have fewer issues on multiple devices right now with CachyOS and Fedora than I do with Windows. If my relative had any capacity for learning new hotkeys, I'd already be wiping his machines and putting Kubuntu LTS on them.
One of my coworker that uses a Mac has an issue with a USB dongle that stopped working randomly, not even with a system update. If he uses on its laptop the external monitor remains blank, the same dongle with another monitor and another identical Macbook with the same macOS version works, the same dongle with the same PC and another brand of monitor works, the same PC with the same monitor and a dongle of a different brand does work, but a different dongle of the same model doesn't. So I guess for some reason macOS saved the ID of that dongle and the setting got corrupted. Tried to reset the hell of it and it doesn't work.
I've never had similar issues with Linux. Most of the time when something doesn't work with Linux you just look at the logs, figure out what is wrong, correct it, and it start work fine again. Configuration of the system is predictable, if you install a software and then you uninstall it the system is left in the same state as before installing it, it doesn't leave files that waste disk space, as well as when you upgrade the system (yes, every Windows or macOS update leaves old stuff behind that it's impossible to delete, effectively "leaking" disk space at the point you have to perform a clean install).
Still some tiny issues with waking from sleep lid closed, but I'm expecting to get that fixed myself with some prompting this week. The people working on Linux on Mac are really impressive (and have access to unlimited tokens), so I expect all issues to be fixed very soon.
Being able combine agents, linux software and Apple hardware is so much fun.
It's been my primary dev machine for 2 years
Mind sharing the provider’s URL?
They're great servers, though, I have a whole fleet of them.
OTOH if you are really concerned about privacy it's probably not the best choice (it's random people's machines).
So any effing claim of being an AI device doesn't end on the demo of best buy laptop where you remove a dog from a photo....
Or that you have "better noise removal" or "you look better on the cam" because that's called noise cancelling and instagram filter and it's being done for ages...
My Strix Halo now can run some models through the NPU and its roughly as fast as its GPU+CPU (on some of the models - some its slower)
I think my 48GB 16 inch MBP will last me 15+ years at this point.
I did actually get a new M5 Pro/48GB machine last month, but that's only because I bought my work machine to keep my personal and professional lives properly separated; otherwise I wasn't seeing any issue running that M3 Max/36GB for a long time further.
One thing that bugs me mentally is leaving that machine pending on my input ("wasted") overnight etc... I'd like to arrive at a workflow where I can have agents running toward the overall goal autonomously overnight as much as they can, and I can review and iterate and "engineer" parts that need it afterwards...
For now, I still guide implementation very closely, piece by piece. Has anyone had any good experiences with workflows / skills / similar for that?
Actually had claude spin up a kanban board to manage my tasks in a 'TODO -> Plan -> Implement -> Review -> QA -> Shipping -> Done' flow (with a worktree for each task) that works great!
https://www.macrumors.com/2026/10/14/apple-three-more-obsole...
https://uptradeit.com/blog/what-macbooks-will-stop-working
https://www.macworld.com/article/673939/this-is-how-long-mac...
Microsoft is known for supporting ancient hardware. Not Apple.
Support for security updates appears to be years longer than Microsoft.
> 48GB 16 inch MBP
That's wicked overkill for ssh + browser. A 15" MacBook Air would work for this really nicely. This is how I use my 13" Air, and for the big screen I dock it at a desk.
I wish they could bring back the 11" Air or 12" MacBook and trim the bezels for something like this. It would be an amazingly portable computer. The 11" iPad Pro with keyboard case is almost this, but you're stuck with iOS and it's actually pretty thick and heavy compared to a MacBook.
* looks inside *
4 kinds of ports. while being generous, and including the SD card slot and the headphone jack.
I don't know anyone who would consider 2-4 to be a "wide array".
It's just handy when you have a USB-A device and don't want a dongle, traveling, etc.
Dongle hell is a real place.
Apple MacBook M5 Max has bandwidth up to 600GB/s, is 170GB/s for M6 the real limit? If so, wouldn't this slow down LLM processing?
Even then, 170GB/s is a bump up from 153GB/s of the M5. So we see a 11% uplift already.
Am I crazy, or are a lot more people using the term "uplift" in place of "increase" within the last year? Google Trends (for whatever it's worth) shows a significant increase (uplift?) since mid-late 2025. [0]
[0] https://trends.google.com/explore?q=uplift&date=today%205-y&...
See also: The use of "learnings" instead of "lessons."
Some people adopt stuff faster than others. I've lived in NZ for nearly 20 years and called someone bro for the first time just last week while playing a sport "should have left it for me bro". Slipped out natural as. Still having bad thoughts about it.
My M2 Pro does 200GB/s.
wtf are we even talking about here.
For a direct comparison to your processor, the M5 Pro is at 308GB/s.
So the theoretical Max would be 680GB/s.
Not that I blame Apple here, for once. Seems like they're trying to cope with the limited quantity of RAM they can procure. And with the paltry bandwidth (170GB/s) of these CPUs they're a bad choice for anything but tiny local models anyway.
They're not making those models for the M6 though; they rerouted their resources to making M7 faster at running local LLMs.
If you want more bandwidth, get an M5 Pro, Max, or Ultra. If you want a lower power base model, get the M6. If you want to have your cake and eat it too, wait for the M7, reportedly in the spring.
Yes, high-capacity LPDDR5X are made and do exist, but for all practical intents and purposes, you can't buy enough of them to deliver to customers at Apple scale.
What was the speed of the base M5?
Because the question is "Which is better for local inference out of the M6 or the M5 Max?"
If an M6 Pro/Max was an option that would change the question.
Also framing AI use for the base M models marketing makes no sense to me. You are way better off getting a used M1 max with 64gb and way higher mem bandwidth when those are the real bottlenecks.
> Apple on Wednesday will announce 14-inch and 16-inch MacBook Pro models featuring touch-enabled OLED displays with thinner bezels, a Dynamic Island instead of a notch, M5 Pro and M5 Max chips, and a lighter and slightly thinner design
https://www.macrumors.com/2026/10/27/macbook-pro-with-touchs...
I must be out of the loop, you skimmed over this. A Mac with a touchscreen??
Is this what we can expect of the new apple? I'm not against AI generated imagery, but I expect them to be of equal quality to a real photo.
Or am I wrong, and it isn't AI generated?
The things that make me think it is just poorly made are 1) the angles of the laptop on the black case, 2) the lack of any lighting on the back panel, 3) the size of the hand on the woman in the back, and the squareness of what she is holding (if it is even supposed to be another mac).
Still a baffling choice.
https://www.apple.com/mac/compare/?modelList=MacBook-Neo-A18...
The base model chip on the Air is binned too - 8 GPU cores on the Air vs 10 on the Pro.
My work device is an M4/32GB. The difference doesnt feel that big, especially when you code mostly with agents.
Is it easily possible to sidegrade/upgrade with a Timemachine backup?
Then do a fresh install on your new system and then use Migration Assistant. Use a USB-4 or a Thunderbolt cable to speed it up.
I checked used prices though, and I think the market is not in our favor. rebuy.de has an M2 Max with 32 GB/1 TB configuration in "very good" condition and is asking 2100 Euro for it.
M6 is "just" 600 Euro more...
I love them all. The M4 Air 14" is from work, so kinda limited in what I can do. The M4 14" Pro is kinda chunky after the Airs. The M2 15" is ideal except on a plane. The M1 13" is great as a personal device on work travel. If only I could not work...
- Orbstack, but just for one containerised service - Laravel Herd for app environments, so all native services instead of containers with bigger VM overhead - Brave browser, profile for work and personal - PHPStorm (quite a hog on RAM) - Now agents in the mix
Had an M1 Pro 8GB but I found 8 to really not be enough. M2 Air 16GB has been a much better sweet spot, albeit it feels like the battery life on it is worse than M1. I've recently started trying to use Zed more and more as my editor instead of PHPStorm and things feel mostly stable, but I still feel like a beefier machine would be good for work purposes.
A bit of googling around seems to point that architecture as somewhat of a problem for others, and might be why Apple quickly moved to put more e-cores on the later generations.
Something to consider. Never managed to get proper distinct confirmation that this was the problem, but a (corporate) laptop upgrade did get me out of this pickle.
Ok, plausible number for that use case. Striking how it seems like the M-series Macs are almost too good enough for most people.
It's almost desperate prose.
(The AI image processing in Pixelmator Pro saved me hours of work on a recent project. Two button clicks vs a day of tweaking. But even that was on my M1 MacBook Air.)
Just the standard chip without Pro/Max makes me assume this is that base one. If so, the price is way high for where that tends to sit, but welcome to buying a computer in 2026 I guess.
It's the same rollout they've used for a couple of generations now. First the base model chips in fall, then the pro/max versions in spring
makes me want to wait another one I have an M1 Max and just set up playing with an M1 (non-max) and realized it can't drive 3 displays. I recall getting Max just for this, but generally, makes me think of non-Max as a non-pro after all.
And having what's presumably the latest CPU but with potentially worse features than the last generation combined with increased prices from RAM is a deal breaker so I'll guess I'll push my M1 for another year.
When I finally cave in and buy something after 5 years I’m very momentarily sad for not perfectly timing the release window, then I realize I keep hardware long enough a half year barely matters.
So Apple can either toss the early production in inventory while the whole process catches up, or they can do what they do. Better for both the company and the consumer that they do the latter.
https://en.wikipedia.org/wiki/Apple_silicon#Comparison_of_M-...
But you are right, they do bin. You can get versions of the cpu with different core counts for this reason.
I’m only seeing this with the pro and max though.
And that should change going forward with smaller chiplets being integrated in package, vs huge Ultra chips.
Compare that with the regilar lineup of other manufacturers: Cryptic model names which are uncomparable and models that are so similar that they could be equivalent.
So, no, 1 is not better but I don't care (as customer) about variety.
I think the Apple experience was better when they focused on delivering a single SKU of each product line, yes. Less decisions as to what to buy, greater focus on making that one SKU stand out.
They’ve always been this way.
The most Apple II models that were sold concurrently were the IIe Platinum, IIc Plus, and IIgs. (And this was around 1990, at which point there were far more models of Macintosh computers on the market.)
64kb RAM + 80col card vs 48kb/40col
The 2e, 2e enhanced, 2c, 2c+ and 2gs were concurrent.
I didn’t bother with Mac because I got tired of writing all the things.
I won't try to defend the 3Gs ;)
Samsung skipped an entire range of numbers from Galaxy S 11 to 19.
Looks like tech companies don't like 9...
And certainly don't look at MikroTik's product naming. ;)
This is “sad”?!? First world problems.
Ryzen AI Max+ Pro 495 or Snapdragon 8 Elite Extreme Gen 6.
Your distaste seems misplaced.
Don't even get me started on USB...
We should have enough distaste for them all
What does a name like Snapdragon 8 Elite Extreme Gen 6 even mean?
- Snapdragon 8 is the highest-end mobile SOC line, as opposed to midrange Snapdragon 7 and low-end Snapdragon 4.
- Elite means that the CPU core is Snapdragon Oryon, and not Arm Cortex off-the-shelf designs.
- Extreme is a trim/upgrade package that gives the GPU more power and performance headroom.
- Gen 6 tells you the generation of chip from that line.
You should refer to prior discussions on this topic if you're genuinely confused, or somehow think that Apple has a superior taxonomy: https://news.ycombinator.com/item?id=50116504
I genuinely do not understand what is "confusing" about this. Like, at all. Annoying maybe? But what is confusing?
the m2/m3 announcements were already kind of cringe to keep saying 'HEY IT'S BETTER THAN INTEL' but at this point it's just embarrassing?
IIRC the Framework 13 Pro beats the MacBook Pro at both video streaming tests and office productivity with 300 bits brightness tests. (Framework live-streamed the video playback test, and the office provided productivity tests I referred to were done by Just Josh Tech)
Perhaps the M6 has leapfrogged but that remains to be seen.
Intel and now Qualcomm are really not far off on performance per watt. Mostly Apple maintains an advantage on single core performance.
The Panther Lake integrated GPU in the X7/X9 SKI is actually faster than the one you get in the base M5.
2. The Intel chips beat out the M5 in multi-core and GPU.
3. These Geekbench scores are close enough that they are in the same performance class. We aren’t talking about 1,000 versus 3,500. The end user isn’t noticing a performance difference between 2700 and 3500, especially since that number is single core only (most workloads can take advantage of multiple threads these days).
Even if we concede that the Intel chip is slower for single core tasks only we can’t just lie and say it’s not in the same performance class. It’s very clearly comparable by objective data.
What we can say definitively by objective measures is that Intel is behind by a year at most, which essentially doesn’t matter in the grand scheme of “which laptop to I buy?” It’s close enough that the processor shouldn’t be part of the buying consideration for the majority of consumers. Intel is not years and years behind what Apple is putting out there, this is not the M1 where Apple was beating Intel in performance and getting double the battery life. We are looking at much more incremental differences where the buyer will not notice a difference over the lifespan of the product.
Here’s a fun one: look up Cyberpunk 2077 benchmarks on Panther Lake compared to M5. It runs natively on both Windows and Mac.
So it's not an apples-to-apples comparison.
probably because of its adjacency to "corpulent" lol
> Mrs. Krabappel: "Embiggens"? I never heard that word before I moved to Springfield.
> Ms. Hoover: I don't know why. It's a perfectly cromulent word.
But now it has made it into the Merriam-Webster dictionary.
Just accept that even if you aren't still using an Intel MacBook Pro, the comparison is still relevant to some part of their target market.
They only just dropped support for Intel Macs from this fall's operating system release (but are still doing security patches for the releases that Intel Macs can run). A year from now, it actually will be fairly reasonable to expect them to give up marketing to the holdouts.
So Apple is now comparing against computers they were selling 7-8 years ago rather than what Intel is delivering now.
Apple didn't fully stop selling Intel macs until June 2023. And they compare against the most recent previous model too.
Source for that fun fact?
Framework 13 Pro, Lenovo ThinkPad X9 15p Aura Edition, for example.
This is of course as of the M5, M6 hasn’t been tested yet.
It is not 2020 anymore. Look up reviews. Just Josh Tech’s ThinkPad X9 15p Aura Edition review, for example.
What you’re saying is just absolutely not true at all and is only backed by your gut feeling.
Further, every Windows laptop seems to have a worse display. For example, an M5 Macbook Pro 14" has a 120Hz 254ppi mini-LED display with a peak brightness of 1600 nits vs. a Dell XPS 14" with a 120Hz 161ppi LCD display with a peak brightness of 500 nits.
So it seems like they have figured out that they can put bigger batteries, worse displays, and throttle TDP to get better battery life, but it is still not possible for them to beat Apple's efficiency just based on those TDP numbers alone—these chips are just drawing more than 2x the power at peak load.
I don't understand why there isn't more effort towards ARM in the Windows world.
As long as they're clear about the comparisons they're making ("X% more battery life than Y" as opposed to simply ("X% more battery life") I'm good with it.
Last time, in 2024, it cost me $6K
Today, it’s $10K
Yowza
Gorilla-arm was a fear based on the naive idea that if you had touch, you would be required to only use touch for every action.
Never understand the appeal of touchscreen on laptop.
Apple software is getting worse every day and while they're not Google or Microsoft they're getting more user hostile at every turn as well.
The lengths they go through to prevent you from doing what you want on their (not your) devices is astonishing.
Very happy with my framework running fedora for many years now... Sure it's not an M1000 but it's almost as good and properly mine
What is it about this MBP that makes it so much more preferable to an MBP with M3 Pro/Max?
But Apple will never sell that to me, as far as I can guess. Apple is not a good person.
It's a company.
15 U.S. Code § 7 - “Person” or “persons” defined
The word “person”, or “persons”, wherever used in sections 1 to 7 of this title shall be deemed to include corporations
https://www.law.cornell.edu/uscode/text/15/7One thing, framing AI use for the base M models marketing makes no sense to me. But honestly this computer is still faster than any of my laptops that came before, even with the visual atrocity of Liquid Glass.
The M6 chip joins M5 Pro and M5 Max to create our most advanced line-up of chips ever built for a pro laptop. With a 12-core CPU, 12-core GPU and our first ever Dual 16‑core Neural Engine, M6 brings next-generation speed and advanced on device AI. All three chips deliver phenomenal single- and multithreaded CPU performance and faster unified memory. . . .The fastest M6 MacBook Pro (24/32GB configs) has a memory bandwidth of 170 GB/s. And the M5 Max is 614 GB/s.
Am I missing some nuance of the M6 MBP here, or am I right to assume it won't set the local LLM world on fire?
M5 Max (128gb) is 600gb/s
Rtx spark laptop (128gb) is 200gb/s.
Desktop 5090 (32gb) is 1700gb/s.
Desktop 6000 (96gb) is 1700gb/s and 600W.
So, yes?
Also, the following two have very different 'needs' to feel good enough:
1. heavy agentic workloads via harness vs 2. a simple private chat interface with a local instance
Even if an MBP's bandwidth were higher, it's tough to see a reason to run it as a constantly loaded 24/7 inference machine due to heat degradation.
P.S. still extremely awesome to be able to host and chat with a decently quantized 27B model on your laptop. ...and to be fair even a 32GB Air would do that! It is a thing of beauty: the amount of folks this may add to the hacker/tinker community.
0. https://float16.cloud/en-en/ai-benchmark/dgx-spark-vs-rtx-50...
There's some value in having a huge amount of memory addressable by the GPU for loading larger models, but even that doesn't make sense for current gen models. For MOE models, newer inference engines with smarter GPU expert caching and pinned memory allow a GPU with lots of bandwidth to significantly outperform a Spark, so that leaves you with larger dense models that fit on a spark but not on a GPU where you might have a use-case.
But even then it's hard. For instance a full-fat BF16 of a ~30B parameter model will fit on a 128GB spark and not a 32GB 5090, but at that size it's going to be frustratingly slow on the spark given the compute and bandwidth constraints.
Imo these devices are mostly good for experimentation, or if you only want to do overnight jobs or training which wouldn't be possible on a GPU, and you don't want to pay $100k for a DGX station.