GPT-5 • Hacker News Board

Re: GPT-5

#141

Post by mlnj » Thu, Aug 07, 2025, 5:20 PM UTC

davepeck wrote: post #121

Sam Altman, in the summer update video: > "[GPT-5] can write an entire computer program from scratch, to help you with whatever you'd like. And we think this idea of software on demand is going to be one of the defining characteristics of the GPT-5 era."

Cannot believe how it could stand up to that high expectation.

But then again, all of this is a hype machine cranked up till the next one needs cranking.

Re: GPT-5

#142

Post by mhh__ » Thu, Aug 07, 2025, 5:20 PM UTC

My conspiracy theory is that the introductory footage of Sam in this and the Jony Ive video is AI generated

Re: GPT-5

#143

Post by arcumaereum » Thu, Aug 07, 2025, 5:20 PM UTC

In terms of raw prose quality, I'm not convinced GPT-5 sounds "less like AI" or "more like a friend". Just count the number of em-dashes. It's become something of a LLM shibboleth.

Re: GPT-5

#144

Post by spruce_tips » Thu, Aug 07, 2025, 5:20 PM UTC

CamelCaseName wrote: post #119

Did they just say they're deprecating all of OpenAI's non-GPT-5 models?

Wonder if deprecating direct access means the gpt5 can still route to those behind the scenes?

Re: GPT-5

#145

Post by andybak » Thu, Aug 07, 2025, 5:21 PM UTC

Ezhik wrote: post #112

I wish the ChatGPT Plus plan had a Claude Code equivalent.

Is that not Codex? Or do you specifically mean the CLI interface?

Re: GPT-5

#146

Post by nicetryguy » Thu, Aug 07, 2025, 5:21 PM UTC

losvedir wrote: post #86

Wait, isn't the Bernoulli effect thing they're demoing now wrong? I thought that was a "common misconception" and wings don't really work by the "longer path" that air takes over the top, and that it was more about angle of attack (which is why planes can fly upside down). It seems like it's actually an ideal "trick" question for an LLM actually, since so much content has been written about it incorrectly. I thought…

Yeah, they sure clicked away from it very fast and kept adjusting the scrollbars. It was confusing what it was trying to display. Furthermore, the prompt contained "Canvas" and "SVG" while as someone with webdev experience these are certainly familiar concepts, i wouldn't consider those in the "casual lexicon" for a random user trying to help a middle schooler with homework. I'm not impressed...

IMO Claude 3.7 could have done a similar / better job with that a year ago.

Re: GPT-5

#147

Post by anonzzzies » Thu, Aug 07, 2025, 5:21 PM UTC

I dont know if there is a faster way to get me riled up: say 'try it' (me a Pro member) and then not getting it because I am logged in. Got opus 4.1 when it appeared. Not sure what is happening here but I am out.

Re: GPT-5

#148

Post by andybak » Thu, Aug 07, 2025, 5:21 PM UTC

jdoe1337halo wrote: post #122

Lmao GPT-5 is still riddled with em dashes. At least we can still identify AI generated text slop for now

You will be foiled by a regex

Re: GPT-5

#149

Post by iSloth » Thu, Aug 07, 2025, 5:21 PM UTC

Wow, they are sunsetting all models after the launch of GPT-5 - Bold statement.

Re: GPT-5

#150

Post by mcs5280 » Thu, Aug 07, 2025, 5:21 PM UTC

mtlynch wrote: post #74

What's going on with their SWE bench graph?[0] GPT-5 non-thinking is labeled 52.8% accuracy, but o3 is shown as a much shorter bar, yet it's labeled 69.1%. And 4o is an identical bar to o3, but it's labeled 30.8%... [0] https://i.postimg.cc/DzkZZLry/y-axis.png

They vibecharted