Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GPT-4.5

openai.com

Post reply on HN
Print view View original

741–750 of 1001 posts

Re: GPT-4.5

#741

Post by einrealist »

acchow wrote: post #379
zaptrem wrote: post #26

GPT 4.5 pricing is insane: Price Input: $75.00 / 1M tokens Cached input: $37.50 / 1M tokens Output: $150.00 / 1M tokens GPT 4o pricing for comparison: Price Input: $2.50 / 1M tokens Cached input: $1.25 / 1M tokens Output: $10.00 / 1M tokens It sounds like it's so expensive and the difference in usefulness is so lacking(?) they're not even gonna keep serving it in the API for long: > GPT‑4.5 is a very large and comput…

> It sounds like it's so expensive and the difference in usefulness is so lacking(?) The claimed hallucination rate is dropping from 61% to 37%. That's a "correct" rate increasing from 29% to 63%. Double the correct rate costs 15x the price? That seems absurd, unless you think about how mistakes compound. Even just 2 steps in and you're comparing a 8.4% correct rate vs 40%. 3 automated steps and it's 2.4% vs 25%.

And remember, with increasing accuracy, the cost of validation goes up (not even linear).

We expect computers to be right. Its a trust problem. Average users will simply trust the results of LLMs and move on without proper validation. And the way the LLMs are trained to mimic human interaction is not helping either. This will reduce overall quality in society.

Its a different thing to work with another human, because there is intention. A human wants to be correct or to mislead me. I am considering this without even thinking about it.

And I don't expect expert models to improve things, unless the problem space is really simple (like checking eggs for anomalies).

Re: GPT-4.5

#742

Post by jtwaleson »

suddenlybananas wrote: post #723
jtwaleson wrote: post #676

Earlier quoted context omitted.

Large parts of my coding are now done by Claude/Cursor. I give it high level tasks and it just does it. It is honestly incredible, and if I would have see this 2 years ago I wouldn't have believed it.

What kind of coding do you do? How much of it is formulaic?

Web app with a VueJS, Typescript frontend and a Rust backend, some Postgres functions and some reasonably complicated algorithms for parsing git history.

Re: GPT-4.5

#743

Post by vbezhenar »

erulabs wrote: post #66

Finally a scaling wall? This is apparently (based on pricing) using about an order of magnitude more compute, and is only maybe 10% more intelligent. Ideally DeepSeeks optimizations help bring the costs way down, but do any AI researchers want to comment on if this changes the overall shape of the scaling curve?

The price is 2x from GPT4. So probably not a decimal order of magnitude.

Re: GPT-4.5

#744

Post by nbzso »

Yesterday I tested Windsurf. Looked the docs and examples. Completed the demo "course" on deeplearning.ai. Gave it the task to build a simple Hugo blog website with a theme link and requirements, it failed consecutive times. With all the available models.

AI art is an abomination. Half of the internet is already filled with AI written crap. Don't start with the video. Soon everyone will require validation to distinguish reality from hallucination (so World ID in place as problem-reaction-solution).

For me, the best use cases are LLM assisted search with limited reasoning. Vision models for digitization and limited code assistance, codebase doc generation and documentation.

Agents are just workflows with more privileges. So where is the revolution? I don't see it.

Where is added value? Making Junior Engineers obsolete? Or make them dumb copy-pasting bio machines?

Depressing a horde of intellectual workers and artists and giving a good excuse for layoffs.

The real value is and always will be in a specialized ML applications.

LLM hype is getting boring.

Re: GPT-4.5

#745

Post by beefnugs »

ur-whale wrote: post #79
zaptrem wrote: post #26

GPT 4.5 pricing is insane: Price Input: $75.00 / 1M tokens Cached input: $37.50 / 1M tokens Output: $150.00 / 1M tokens GPT 4o pricing for comparison: Price Input: $2.50 / 1M tokens Cached input: $1.25 / 1M tokens Output: $10.00 / 1M tokens It sounds like it's so expensive and the difference in usefulness is so lacking(?) they're not even gonna keep serving it in the API for long: > GPT‑4.5 is a very large and comput…

AI as it stands in 2025 is an amazing technology, but it is not a product at all . As a result, OpenAI simply does not have a business model, even if they are trying to convince the world that they do. My bet is that they're currently burning through other people's capital at an amazing rate, but that they are light-years from profitability They are also being chased by fierce competition and OpenSource which is very…

Yes: the real truth is, if there really was a good AI created, then we wouldnt even know about it existing until a billion dollar company takes over some industry with only a handful of developers in the entire company. Only then would hints spill out into the world that its possible.

No "good" AI will ever be open to everyone and relatively cheap, this is the same phenomenon as "how to get rich" books

Re: GPT-4.5

#746

Post by ta3456345345 »

The AI hyperbole is so cringe right now (and for the last few years). I've yet to see anyone come up with something that'd wow me, and say, "OK, yep, that deserves those cycles".

Writing terrible fanfic esque books, sometimes OK images, chatbot style talking. meh.

Re: GPT-4.5

#747

Post by PeterStuer »

Currently my daily API costs for 4o are low enough and performance/quality for my usecases good enough that switching models has not made to to the top of application improvements.

My cases' costs are more heavily slanted towards input tokens, so trying 4.5 would raise my costs over 25x, which is a non-starter.

Re: GPT-4.5

#748

Post by jsemrau »

PeterStuer wrote: post #747

Currently my daily API costs for 4o are low enough and performance/quality for my usecases good enough that switching models has not made to to the top of application improvements. My cases' costs are more heavily slanted towards input tokens, so trying 4.5 would raise my costs over 25x, which is a non-starter.

Interesting observation.It seems capability has reached a plateau. Like a local maximum.

Re: GPT-4.5

#749

Post by _1tem »

killerstorm wrote: post #736
erulabs wrote: post #66

Finally a scaling wall? This is apparently (based on pricing) using about an order of magnitude more compute, and is only maybe 10% more intelligent. Ideally DeepSeeks optimizations help bring the costs way down, but do any AI researchers want to comment on if this changes the overall shape of the scaling curve?

It depends on how you compare. On a subset of tasks I'm interested in, it's 10x more intelligent than GPT-4. (Note that GPT-4 was in many ways better than 4o.) It's not a coding champion, but it knows A LOT of stuff, excellent common sense, top quality writing. For me it's like "deep research lite". I found OpenAI Deep research excellent, but GPT-4.5 might in many cases beat it.

> On a subset of tasks I'm interested in, it's 10x more intelligent than GPT-4.

Very intriguing. Care to share an example?

Re: GPT-4.5

#750

Post by j_maffe »

chippiewill wrote: post #575
j_bum wrote: post #518

Earlier quoted context omitted.

Any examples you’d be willing to share?

They have examples in the announcement post. It does a better job of understanding intent in the question which helps it give an informal rather than essay style response where appropriate.

I wouldn't call that "too insane." As others have pointed out, you can get similar results from fine-tuning the RLHF.
Post reply on HN