GPT-5 • Hacker News Board

Re: GPT-5

#81

Post by octo888 » Thu, Aug 07, 2025, 5:11 PM UTC

atonse wrote: post #2

For day to day coding, I've found Anthropic to be killing it with Sonnet 3.7 and now Sonnet 4, and Claude Code feeling like it has even bigger advantages over when it's used in Cursor (And I can't explain why). I don't even try to use the OpenAI models because it's felt like night and day. Hopefully GPT-5 helps them catch up. Although I'm sure there are 100 people that have their own personal "hopefully GPT-5 fixes m…

Killing it - at what type of coding task? What "bigger advantages" specifically? What is night and day?

Re: GPT-5

#82

Post by bigyabai » Thu, Aug 07, 2025, 5:11 PM UTC

Workaccount2 wrote: post #64

OpenAI taking a page out of Apple's book and only comparing against themselves

Presumably because GLM 4.5 or Qwen3 comparisons would clobber them in eval scores.

Re: GPT-5

#83

Post by clueless » Thu, Aug 07, 2025, 5:11 PM UTC

bmau5 wrote: post #36

It's very interesting how memetic the language around different models is. Elon seems to have coined "PhD level intelligence in all topics" and now Sam repeated it in his presentation. Despite it not having an actual meaning. I think OpenAI will coin they've achieved AGI first (as they have incentives to based on the rumored contract with MSFT), and then everyone will claim we've achieved it.

Elon did not coin this, Kurzweil has been using this coinage for a lot longer.

Re: GPT-5

#84

Post by croemer » Thu, Aug 07, 2025, 5:11 PM UTC

The presentation asks for a moving svg to illustrate Bernoulli, that's suspiciously close to a Pelican.

Re: GPT-5

#85

Post by koolala » Thu, Aug 07, 2025, 5:11 PM UTC

xyst wrote: post #12

> comments turned off yikes - the poor executive leadership’s fragile egos cannot take the criticism.

What's up with their very first eval? The SWE bars and numbers don't line up.

Re: GPT-5

#86

Post by losvedir » Thu, Aug 07, 2025, 5:12 PM UTC

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 at first they were going to demo this to show that it knew better, but it seems like it's just regurgitating the same misleading stuff. So, not a good look.

Re: GPT-5

#87

Post by arm32 » Thu, Aug 07, 2025, 5:12 PM UTC

RivieraKid wrote: post #54

Is it bad that I hope it's not a significant improvement in coding?

What's bad about not wanting to lose your job?

Re: GPT-5

#88

Post by maltsev » Thu, Aug 07, 2025, 5:12 PM UTC

gpt-5 is now #1 at LMArena: https://lmarena.ai/leaderboard/text

Re: GPT-5

#89

Post by b800h » Thu, Aug 07, 2025, 5:12 PM UTC

This livestream is atrocious

Re: GPT-5

#90

Post by nonhaver » Thu, Aug 07, 2025, 5:12 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

also wondering this. had to pause the livestream to make sure i wasnt crazy. definitely eyebrow raising