[gehsty]

This kind of thing just makes me think Apple will get to a point where they have good enough local models and good enough harnesses for doing things, and most normal people will just use them… Does the LLM become another interface to computing?

[ibero]

i agree.

i believe that for most people on the street, for most tasks, a Chat GPT 3.5 era LLM is sufficient enough. sprinkle in tool calling and other things, and that becomes enough. if you can prioritize that level of a model on-device (baking it in etc), then you can bifurcate AI users between those unwilling to pay and those who are willing to pay A LOT for frontier model performance.

[trvz]

Anyone even a little bit serious will be paying either way. If not for the cloud models, then for hardware upgrades.

[fy20]

Gemma 4 is already closer to gpt-4o level of intelligence, however it's knowledge is much lower.

[SOLAR_FIELDS]

This question hinges on whether model advancement plateaus enough for machine sized models to compare to frontier performance. If it does, the answer is yes. If it doesn’t, the answer is no

[jfaat]

I disagree. The point of the frontier models is to do everything as well as possible ("AGI" race or whatever) but smaller models with some RL are going to be the clear winner for a ton of use cases. Think about all the use cases for LLMs that would never be economical at frontier inference costs, and in no way need it. You don't need or even want a phd polymath helping you with small productivity tasks that most people use computers for every day. It's often overwrought and annoying. I don't even really like the frontier models for coding for this reason. They're constantly blowing up scope and you have to fight it constantly.

[cptskippy]

More likely, it's going to be whether frontier models advance enough that most people would be willing to pay for them. Right now they don't, but a model you can run locally for free on hardware you already own is very compelling because, while they're not as good as Frontier Models, they're still pretty good.

Tools like Opencode demonstrate that when you box them in tightly enough they can actually be pretty competent.

[c7b]

Neural machines were always going to be an alternative computing paradigm to von Neumann machines. Had it not been for Minsky we would arguably have gotten to a point where they're useful sooner. But why do you say that as if it's a small thing?

[ls612]

I have thought this for a while. Computing 1.0 meant that we needed to learn the computer’s language to interact with the computer fully. Computing 2.0 is that now the computer has learned our language instead.

[bbg2401]

I’m not sure what interacting with the computer fully means, but I don’t think it’s true about most computer users. They use software which invariably speaks to the user through visual metaphors and conceptual abstractions. Nothing has changed in that regard. This is true even of most programmers using high level languages.

Nothing has changed at the level of computation either, hardware speaks the same vocabulary and of electrical signals as ever.

Could you describe the change you are observing?

[throw1234567891]

People don’t type so much anymore.

[bigyabai]

Why wait? People are already doing their work on OpenAI and Anthropic's servers, Apple Intelligence servers could quickly subsume any "local" model work that you want to do.

That way everyone has access, even with older devices, and it's a subscription! Then Apple can tie their APIs into the ecosystem you love at a flat cost you can afford. No need to support local model integration in the first place, problem solved.

[mark_l_watson]

Apple’s System Model is actually pretty good but is hard-limited to a 4K context length. This is OK for small Python utilities that need a model for applications that operate on small amounts of data, but is a disappointing limitation.

That said, and this is off topic: Siri on the newest iOS beta is surprisingly good now. I asked it what model it was using yesterday and it said Gemini for difficult problems, then secure Apple model in cloud, and local Apple model.

[jjice]

Looks like iOS 27 ships with a 20B sparse model for on-device work, so I wonder if that comes with a context length bump. They're still pushing for better local models, which is a good sign for the future for their products.

https://runaihome.com/blog/wwdc-2026-apple-ai-home-lab-verdi...

[dominotw]

or the other way where primary interfaces ppl use computing arent apple devices like laptops and phones.

[chickensong]

> Does the LLM become another interface to computing?

It already is.

[gehsty]

It literally is, but people don’t use LLMs to interact with their computers they use touch screens or mouse and a keyboard.

This is why I think it will be the platform owners who end up winning - if iPhone / Mac / iPad ship with a good enough local model that can directly interact with the OS and apps and the web people might actually start talking to their computers.

[chickensong]

A portion of the leading edge is already talking to their computers. First it was wiring up Wispr to your agent, now you can just enable voice mode in Claude. That paradigm shift just hasn't reached the mainstream yet.

Most people can barely type or operate a computer, so I think it will be a boon to many folks to have a voice/agent interface to their computing. I agree that a local model that's "good enough" will suffice for most general tasks. There's no need to call Mythos to "add cheese to the shopping list" or "summarize today's emails". Harder tasks will call out to stronger models, but many people just won't need that power.

I suspect all the platforms are keenly aware of this, and Apple will drop Siri v2 or whatever when they're satisfied with their UX and more confident the agent won't go off the rails.

[doctorpangloss]

based on their apple intelligence demos they are optimizing their products for their core demo of 55-95 year old boomers who talk out loud to think and read every page of the nytimes. you are miles away from the US consumer product experience here.

[gehsty]

You say this like it a bad thing my 90yr old grandmother can use WhatsApp and iCloud Photos - if they can get that demographic using LLMs meaningfully, it probably means they’ve cracked the interface.

[marcus_holmes]

Boomers were born up until 1965, so the youngest boomer is 61 this year.

Please don't slur us older GenX as boomers.