[questinthrow]

Question, what stops openai from downgrading existing models so that you're pushed up the subscription tiers to ever more expensive models? I'd imagine they're currently losing a ton of money supplying everyone with decent models with a ton of compute behind them because they want us to become addicted to using them right? The fact that classic free web searching is becoming diluted by low quality AI content will make us rely on these LLMs almost exclusively in a few years or so. Am I seeing this wrong?

[jjice]

It's definitely not impossible. I think the increase competition they've begun to face over the last year is helping as a deterrent. If people notice GPT 4 sucks now and they can get Claude 3.5 Sonnet for the same price, they'll move. If the user doesn't care enough to move, they weren't going to upgrade anyway.

[SoftTalker]

Also depends on the friction to move. I admittedly have not really started using AI in my work, so I don't know. Is it easy to replace GPT with Claude or do I have to reconfigure a bunch of integration and learn new usage?

[liamYC]

It depends on the tool you use and I guess the use case too. Some are language model agnostic like aider in the command line, I use sonnit sometimes and then 4o other times. I wonder if or when language models will become highly differentiable. Right now I see them more like a commodity that are relatively interchangeable but that is shifting slightly with other features as they battle to become platforms

[drdrey]

competition is what stops them from downgrading the existing stuff

[turblety]

and is also exclusively the reason why Sam Altman is lying to governments about safety risks, so he can regulate out his competition.

[JanisErdmanis]

They don’t need to downgrade what is already downgraded. In my experience ChatGPT was much more capable a year ago than it is now and have become more dogmatic. Their latest updates have focused on optimizing benchmark scenarios while reducing computation costs.

[mtmail]

> I'd imagine they're currently losing a ton of money supplying everyone

I can't tell how much they loose but they also have decent revenue "The company's annualized revenue topped $1.6 billion in December [2023]" https://www.reuters.com/technology/openai-hits-2-bln-revenue...

[swores]

What's important, and I don't think has ever been revealed by OpenAI, is what the margin is on actual use of the models.

If they're losing money but just because they're investing billions in R&D, while only spending a few hundred million to serve the use that's bringing in $1.6B then it would be a positive story despite the technical loss, just like Amazon's years if aggressive growth at the cost of profits.

But if they're losing money because the server costs needed for the use that brings in $1.6B are $3B then they've got a scaling problem until they either raise prices or lower costs or both.

[derac]

competition?

[adamtaylor_13]

Oh, you mean what they did with GPT-4 to make o1 look better and then push everyone to anthropic?

Eh… probably everyone moving to anthropic.