Mostly agreed, however I'm not sure about 3: I suspect it works like gym memberships, and the companies mostly make their money from people who don't use the subscriptions all that much.

▲

FinnLobsien 11 hours ago | parent | next [-]

I think the problem is that the companies mostly don't make money, period. They may have better unit economics on underused subscriptions, but I don't see a world in which OAI/Anthropic don't heavily tighten the screws in the future.

Right now it's silly to default to frontier models, but it won't bankrupt your company. I believe in the short-medium term future, we'll need to be more deliberate about model choices.

In the long-term, of course, tech costs tend to plummet. Is there a future where in 15 years, my Apple Watch locally runs an Opus 4.8-class model? Maybe. And that would obviate this whole discussion.

▲

hparadiz 9 hours ago | parent [-]

I'm just here so I can look at my post history and have a hearty laugh about in a few years.

▲

piva00 8 hours ago | parent [-]

You can save it to your favourites, no need to comment at all if it's not going to add to the conversation.

▲

hparadiz 7 hours ago | parent [-]

Okay here is my adding to the conversation:

The current discourse about LLMs in coding especially is based on the cheapest type of inference: text. This technology was designed for images which is a much more computationally expensive task than text. If it's already profitable to use this technology for multimedia like images and videos then using it on a text based inference for code is less then 1% as computationally expensive. Furthermore in the aggregate over time the computational expensive of text based inference precipitates negatively. In other words using it to write code will inevitably become a throwaway computational task like decompressing a jpeg. And yes decompressing jpegs would lag your 386 in the early 90s.

▲

piva00 6 hours ago | parent [-]

What?

LLMs were designed for text, it's in their name "large language model". Only with specialised encoders like vision transformers they were able to process images as well but you're absolutely wrong about the original design intent.

In the end you just added misinformation, just save the comment to your favourites and set a reminder to check it again in a few years like you wanted.

▲

hparadiz 6 hours ago | parent [-]

The first technological breakthroughs were with face and red eye detection in 2003. Then object detection between 2008-2012. Text models didn't become useful until about 2016. Please watch the first course of Dr Fei Fei Li's lectures on the subject.

	▲	piva00 3 hours ago \| parent [-]
		If we want to keep tracing the lineage of AI we'll have to go all the way back to Markov chains from the 70s. You said LLMs were designed for images which is absolutely incorrect.

▲

Cthulhu_ 10 hours ago | parent | prev | next [-]

I'd say that is/was their long game, but it's still very much in hype phase so there's a lot of people intensively using these models, and I don't think it's anywhere near cost efficient right now. Maybe in the long run when people get bored with it, but on the other hand people are becoming dependent on it for everyday things.

We've already seen price hikes / token limits earlier this year, with suddenly some people running out of budget on the first day of the month. This will likely keep going for a while.

On the other hand, costs will drop too - open models and specialized hardware, as the article notes. The long question will be whether the companies will get a return on their invested billions. I don't think they will, not with the amount of competition they're facing, and I don't think any one company or model (series) has a monopoly yet. Popularity sure, but I'm confident a competitor may appear tomorrow and people will switch.

▲

iamacyborg 11 hours ago | parent | prev | next [-]

I follow a guy called Daniel McCarthy on LinkedIn who writes a lot on CLV and that seems to be his take. Even if theoretically you get way more than you pay with subscriptions, the vast majority of people are not power users.

https://danielminhmccarthy.com/

▲

dofm 11 hours ago | parent [-]

The vast majority of active users of ChatGPT could successfully use a model like Gemma 4 12B with agentic search if x86 hardware didn't make that so difficult.

Likely even the E4B, which is really both fun and impressive.

That is clearly a big component of Apple's bet, anyway.

	▲	avadodin 10 hours ago \| parent [-]
		I have experimented with it and E4b is perfectly capable of being useful if you provide it with ready–to–use skills. It's still more like programming than telling a chatbot to go make you GTAVI in JavaScript and make sure the graphics are as good as the original. Maybe a safer prediction would be that most people will be fine just using hybrid agentic programs that run the models locally(probably with extra spyware). I think this is Apple's bet.

▲

xienze 10 hours ago | parent | prev | next [-]

> I suspect it works like gym memberships, and the companies mostly make their money from people who don't use the subscriptions all that much.

I think it's like that, but not quite. The people who have a subscription but barely use it were probably never doing any serious work with AI in the first place. I.e., why would they get a subscription when their one or two chat questions (or, "make a picture of me as a superhero" prompts) per day can be had for free?

Especially with Claude, I think people who subscribe skew very heavily towards people that can very easily make more than $20 worth of queries in a month. And then there's the not-insignificant number of people who are tokenmaxxing.

It's like the gym membership model except ten percent of members are able to spend 72 hours per day at the gym while the rest spend 8 IMO.

	▲	usef- 9 hours ago \| parent [-]
		Based on the people I know, they're paying because when ask they want the smartest model to be the one answering. There's still quite a difference between models.

▲

PunchyHamster 10 hours ago | parent | prev [-]

Technically yes but it's not hard to get to $20 plan caps. Till current hardware prices cool down I don't see it being easy to make money on frontier models.