> Available on iPhone 17 Pro, iPhone 17 Pro Max, iPhone Air, iPad models with M4 and later and at least 12GB of unified memory, and Mac models with M3 and later and at least 12GB of unified memory.
It’s really disappointing to see the on-device models being limited to so few devices. And this was after the iPhone 16 and 16 Pro were marketed so heavily with supporting their now failed effort at AI.
Apple has pulled a Tesla here. FSD on HW3 cars is stuck on old software with no upgrade path as of now. Tesla is potentially justifying it by calling it "FSD (Supervised)" so they don't have to do an expensive retrofit to them even though they sold these cars originally with the promise of fully autonomous driving.
All the iPhone 16/Pro owners have been waiting for Apple Intelligence features announced from that WWDC 2 years ago. They didn't get delivered and now won't ever be delivered with on-device intelligence due to the 8GB RAM limitation.
I’ve got a 2023 Mac Studio M2, and was dismayed by the M3 & later. So I’ve been trying to track down more details. That specific device list is only for:
> Apple’s most powerful on-device model and the features it enables, like expressive voices and more advanced dictation, […]
On other devices, I think there’s still on device support (just not with the “most powerful model”), for these devices:
> Apple Intelligence and Siri AI in iOS 27, iPadOS 27, macOS 27, watchOS 27, and visionOS 27 are available on iPhone 16 models or later, iPhone 15 Pro, iPhone 15 Pro Max, iPad mini (A17 Pro), MacBook Neo (A18 Pro), iPad models with M1 or later, Mac with M1 or later, Apple Vision Pro, Apple Watch Series 9 or later, Apple Watch Ultra 2 or later, and Apple Watch SE 3 when paired with an Apple Intelligence-enabled iPhone nearby.
That's specifically the secondary, more-powerful model. It was only mentioned in passing in the keynote, but on this page anyway, it seems to be just the improved dictation in Siri and ability to customize pacing, etc.
I'm glad I've been sitting on my iPhone 15 Pro Max... I'll upgrade someday, if/when I need to and the software updates are compelling, but I'll see how things run on my M5 Macbook. But the 15 Pro Max isn't subpar in any other way.
> It’s really disappointing to see the on-device models being limited to so few devices.
At first I thought it was the usual planned obsolescence. Then I realized it may be a true technical limitation. I suspect an embedding model is required to run on device in order to make several of the features work. Embedding models are small compared to LLMs, but, depending on their capabilities, could be the memory driver.
> At first I thought it was the usual planned obsolescence. Then I realized it may be a true technical limitation.
It's shortsightedness on Apple's part, being both the first to the game, and the last to the game. Go back any number of years in the past decade. For at ;east 10 years Apple was telling everyone how great their local models are, and how amazing desktop-class their "neural" chips are. And yet here we are with them somehow incapable of running anything except on a few recent top-tier models.
As a fellow fullscreen liker (there are dozens of us! dozens!), this looks quite intriguing to me. A grid layout always fit better with my mental model of how these types of spaces should work, since I could use rows as categories of work and columns as specific applications within that category. Or one of a few other mental models I've used over the years.
This behavior from people is the one thing that makes me wonder if we all wouldn't be better off just chucking AI off the proverbial cliff. It should be useful tool for enhancing the tasks we have to do, not something to fully replace thinking and human interaction completely.
The problem for me is that this is littering my drive with files in directory's for browsers I have never installed. I have never installed Arc or Brave or a few other of the browsers, but now I have directory's for them in my Application Support folder.
Is it an easy cleanup? Sure. But I shouldn't be seeing support folders for apps I have never installed on my machine, ever.
> But I shouldn't be seeing support folders for apps I have never installed on my machine, ever.
I don't know why you think this?
Macos uses application support as a place for user preferences and configuration. That chrome directory claude desktop creates is configuration FOR CLAUDE DESKTOP that instructs chrome (existing or future installs) that it's allowed to start processes related to claude desktop.
On windows - this would have been a registry key for chrome pointing to the manifest file location
On linux - it would be in .config/google-chrome
These directories/settings explicitly exist in well-known locations to allow applications to configure and interoperate with each other.
---
That fact that you might not have chrome installed at this moment is irrelevant. Honestly, I much prefer that it place the file regardless for two reasons
1. You might install chrome later, and this flow shouldn't require you to start claude desktop again to work
2. To be smart here is WAY worse. It would mean claude desktop needs to scan your machine for a variety of applications on each launch, get their install locations correct across each OS version, and have permissions to do this check, and then change behavior based on the results. Which I imagine based on the hyperbole in this thread already... is going to get them even worse press - not to mention be just incredibly annoying to maintain and flakey to support.
Tablets need an edge where you can grip it. Without thicker bezels, it’s harder to hold it without your fingers being on the screen. This is much less of an issue for phone-sized devices.
- Why is grip a feature of the bare tablet and not part of a case accessory?
- Why is the grip point the flat glass front of the display, instead of anything more ergonomic for actually holding it?
Phones don't do this, not even 7" phablets, nor for holding them horizontally, nor holding them with two hands gamepad-style during gameply. Why do tablets?
Because most tablets are intended to be as thin and compact as possible while being too large to wrap a hand around. Imagine the complaints if Apple told people to buy a case so they could hold the product. Imagine putting a ledge on one side to hold it oh hey, it's a Kindle Scribe (and still ALSO has a bigger bezel than the iPad Pro.)
Unfortunately how I feel about it too. I gave an honest effort at getting into the ecosystem and tested it out with a few close friends. The rough edges brought the experience down compared to other stuff that “just works”, and losing community support for the IRC bridge took a huge use of my own away from it.
The first time I went through the theme picker, I was tickled to see a theme I had made years ago included. Realized later it was due to it including all of the themes from iTerm2 Color Schemes automatically.
It made for a more fun first experience with a terminal emulator than I expected to have.
Where is it clear that there is great market demand? The only time I see devices like this discussed is in the small maker community. And the technology limitation of not being a fast updating display limits it to more niche uses.
Half of all supermarkets in Finland have e-ink screens for product price tags along shelves. It saves tons of printing. In a country full of trees. Must be great value.
I think the primary savings of e-ink pricetags is in labor, not materials. Being able to just send price changes directly to the shelf without having humans go through and change them out is a big deal, so to speak.
Also less error prone and more importantly much faster.
Being able to update prices in near real time would also potentially allow for more efficiency in setting prices, meaning you can set them in real time.
With real time stock information, you could handle sudden demand spikes much more gracefully. Have you prices automatically adjusted between an minimum and maximum price based on current stock and demand. Basically get you targeted stock turnover for the best possible price. (Best coupled with a self-checkout system so the customer consents to the current price).
Not that any supermarket I know of is currently smart enough to do that but they could. With all the hoarding due to shortages I could see them smartening up one day.
> Not that any supermarket I know of is currently smart enough to do that but they could
This may or may not be illegal as fuck. Picking up an item that costs 1$ and having it cost 1.2$ by the time you get to the checkout is all sorts of wrong.
The Price Accuracy Policy allows customers to be compensated in case of a pricing error at the register. If the price of the article you are buying is higher at the register than what was shown on the shelf, the merchant must:
give you the item for free if the item costs less than $10;
sell you the item at the shelf price, minus $10, if the item costs more than $10.
So pricing could change at store close, but it'd be risky to change otherwise. And of course, it should* be a problem. You pick up a product, get to cash, and the price sneakily changes and you don't notice? Uncool.
Paper price have to be changed every week for specials even in Quebec.. it is very labour intensive and very error probe. They also often fall off. Eink price displays are meant to help that.
I used to work in a groceries store and changing price used to take so long that we had to do even while opened with shopping customers..
There's a whole slew of ways to scam retail prices. Sometimes customers used to just peel a price tag off one product and place it on another. (Store employees used to, less than 40 years ago, put a price sticker manually on every box. Then they'd key the price into the register.) Sometimes if there's a SKU sticker rather than the preprinted UPC sticker on the package from the manufacturer people will still try this.
This is a posted price, marked by the store. They'd need to forge it, the logo, font, etc.
That said, there are usually security cameras, and that would be fraud.
People shoplift too, so it is a similar type of crime.
You're assuming sub-day changes. Stock doesn't necessarily come in every day and responding to demand spikes on a day by day basis solves this for any non-24h shop.
You could also do sub-day lowering of prices, although I think day by day changes adds a lot on its own if you want to have dynamic prices in response to demand.
I think there are some self-checkout solutions where you scan the barcode while taking the product off the shelf instead of at the end. The price should, of course, not change between picking up and payment.
Automatic pricing updates on shelving displays have been around for awhile, eink versions have gotten more popular because they use so much less power. That combined with low power bluetooth has made it even easier to roll them out.
I live in Kazakhstan which is country without trees. I never saw eink price tags. They're paper. I doubt it has anything with trees. Paper is too cheap.
Sure they are eInk? On Carrefour here they have had plain old LCD ones for ages now. It looks like e-ink, but on close inspection it is definitely not (e.g. mirrorish background color). Polarized glasses also help detection :)
And that is what I do agree e-Ink is overrated. People have trouble distinguishing it from reflective LCDs (remember the Pebble).
On the flip side, they're doing away with loot boxes in Overwatch 2 and replacing it with a battlepass and straight-forward shop[1]. Plenty of opportunity to screw that up, of course, but at least it'll be easier to see than having to dive into drop rate algorithms and best guesses at such.
I liked the formula of having a price and then loot boxes. This is because they had to make the lootboxes content available eventually, which it is what happened: I have all the skins I like and I haven't spent a dime for them.
Being free + bp they can lock all cosmetings behind $$$ and push even more fomo tactings to your face
It’s really disappointing to see the on-device models being limited to so few devices. And this was after the iPhone 16 and 16 Pro were marketed so heavily with supporting their now failed effort at AI.