GPT-5.4 • Hacker News Board

Re: GPT-5.4

#61

Post by nthypes » Thu, Mar 05, 2026, 6:41 PM UTC

nthypes wrote: post #43

$30/M Input and $180/M Output Tokens is nuts. Ridiculous expensive for not that great bump on intelligence when compared to other models.

Gemini 3.1 Pro

$2/M Input Tokens $15/M Output Tokens

Claude Opus 4.6

$5/M Input Tokens $25/M Output Tokens

Re: GPT-5.4

#62

Post by rvz » Thu, Mar 05, 2026, 6:42 PM UTC

nthypes wrote: post #43

$30/M Input and $180/M Output Tokens is nuts. Ridiculous expensive for not that great bump on intelligence when compared to other models.

You didn't realize they can increase / change prices for intelligence?

This should not be shocking.

Re: GPT-5.4

#63

Post by GenerWork » Thu, Mar 05, 2026, 6:42 PM UTC

elliotbnvl wrote: post #39

denysvitali wrote: post #6

Article: https://openai.com/index/introducing-gpt-5-4/ gpt-5.4 Input: $2.50 /M tokens Cached: $0.25 /M tokens Output: $15 /M tokens --- gpt-5.4-pro Input: $30 /M tokens Output: $180 /M tokens Wtf

Looks like it's an order of magnitude off. Missprint?

Looks like an extra zero was added?

Re: GPT-5.4

#64

Post by paxys » Thu, Mar 05, 2026, 6:42 PM UTC

cj wrote: post #50

I use ChatGPT primarily for health related prompts. Looking at bloodwork, playing doctor for diagnosing minor aches/pains from weightlifting, etc. Interesting, the "Health" category seems to report worse performance compared to 5.2.

Models are being neutered for questions related to law, health etc. for liability reasons.

Re: GPT-5.4

#65

Post by observationist » Thu, Mar 05, 2026, 6:42 PM UTC

chabes wrote: post #21

twtw99 wrote: post #15

If you don't want to click in, easy comparison with other 2 frontier models - https://x.com/OpenAI/status/2029620619743219811?s=20

Definitely don’t want to click in at x either.

[flagged]

Re: GPT-5.4

#66

Post by timpera » Thu, Mar 05, 2026, 6:42 PM UTC

> Steerability: Similarly to how Codex outlines its approach when it starts working, GPT‑5.4 Thinking in ChatGPT will now outline its work with a preamble for longer, more complex queries. You can also add instructions or adjust its direction mid-response.

This was definitely missing before, and a frustrating difference when switching between ChatGPT and Codex. Great addition.

Re: GPT-5.4

#67

Post by minimaxir » Thu, Mar 05, 2026, 6:42 PM UTC

wahnfrieden wrote: post #58

No Codex model yet

GPT-5.4 is the new Codex model.

Re: GPT-5.4

#68

Post by jstummbillig » Thu, Mar 05, 2026, 6:42 PM UTC

mattas wrote: post #5

"GPT‑5.4 interprets screenshots of a browser interface and interacts with UI elements through coordinate-based clicking to send emails and schedule a calendar event." They show an example of 5.4 clicking around in Gmail to send an email. I still think this is the wrong interface to be interacting with the internet. Why not use Gmail APIs? No need to do any screenshot interpretation or coordinate-based clicking.

Because the web and software more generally if full of not APIs and you do, in fact, need the clicking to work to make agents work generally

Re: GPT-5.4

#69

Post by yanis_t » Thu, Mar 05, 2026, 6:42 PM UTC

These releases are lacking something. Yes, they optimised for benchmarks, but it’s just not all that impressive anymore. It is time for a product, not for a marginally improved model.

Re: GPT-5.4

#70

Post by steve1977 » Thu, Mar 05, 2026, 6:42 PM UTC

BoredPositron wrote: post #60

steve1977 wrote: post #42

Earlier quoted context omitted.

One could argue that LLMs learning programming languages made for humans (i.e. most of them) is using the wrong interface as well. Why not use machine code?

because they are inherently text based as is code?

But they are abstractions made to cater to human weaknesses.