OpenAI announces GPT-5

I WANT IT I NEED IT
 
burning-money-finance.gif


By all means, keep going.
 
this one is a bit underwhelming, particularly with everything else going on in the AI space

Probably the one nice convenience feature they mentioned for me was the ability to supply custom context-free grammars (basically syntax rules) to be applied to custom tool calls. You can already do this with tools like llama.cpp locally on a ton of models, but API support on large models was absent.
 
Benchmarks are solid, but nothing really mind-blowing. But honestly I stopped paying much attention to benchmarks quite some time ago, I just use the models myself and evaluate accordingly. Drastic reduction in hallucination is a big deal and very welcome. Immediate rollout to the free tier is a pretty good deal, too bad I renewed my Plus annual literally three days ago.

Anyway, this was a pretty weird presentation. No real bangers, quite a lot of filler (why?) and I don't really feel the OpenAI "San Francisco youngster geek squad" vibe, and the cancer lady was weird and unnecessary. Sam Altman is just a charismatic void, a black hole personality. Also nothing on agent systems?

I dunno, I expected a banger but maybe we've reached a point of diminishing returns with these large language models in their current iteration.

99% of normies are not going to notice a difference between 4o and 5.

Edit: and they're getting clowned on for their bizarre graphs.



 
Last edited:
Benchmarks are solid, but nothing really mind-blowing. But honestly I stopped paying much attention to benchmarks quite some time ago, I just use the models myself and evaluate accordingly. Drastic reduction in hallucination is a big deal and very welcome. Immediate rollout to the free tier is a pretty good deal, too bad I renewed my Plus annual literally three days ago.

Anyway, this was a pretty weird presentation. No real bangers, quite a lot of filler (why?) and I don't really feel the OpenAI "San Francisco youngster geek squad" vibe, and the cancer lady was weird and unnecessary. Sam Altman is just a charismatic void, a black hole personality. Also nothing on agent systems?

I dunno, I expected a banger but maybe we've reached a point of diminishing returns with these large language models in their current iteration.

99% of normies are not going to notice a difference between 4o and 5.

Edit: and they're getting clowned on for their bizarre graphs.




Did Chat GPT hallucinate those charts?
 
Each day I pray that the Eye of Sauron that is AI development does not turn to my job function.

It's only a matter of time until I'm automated out. Luckily they seem to be focusing on entry-level positions and traditional programming for now.
 
Each day I pray that the Eye of Sauron that is AI development does not turn to my job function.

It's only a matter of time until I'm automated out. Luckily they seem to be focusing on entry-level positions and traditional programming for now.
My wife is pretty well paid in the healthcare field. Worst case scenario... I become a stay-at-home dad with a side gig as a welder or something.
 
Each day I pray that the Eye of Sauron that is AI development does not turn to my job function.

It's only a matter of time until I'm automated out. Luckily they seem to be focusing on entry-level positions and traditional programming for now.

Might want to consider a career change then.

Sitting around doing nothing while you know the end is coming sooner rather than later is going to fuck you up when the time comes.
 
Might want to consider a career change then.

Sitting around doing nothing while you know the end is coming sooner rather than later is going to fuck you up when the time comes.
I appreciate you looking out and giving advice, but please don't assume I'm doing nothing to prepare for what I think is inevitable.
 
Might want to consider a career change then.

Sitting around doing nothing while you know the end is coming sooner rather than later is going to fuck you up when the time comes.

I appreciate you looking out and giving advice, but please don't assume I'm doing nothing to prepare for what I think is inevitable.

would be fun to have a thread for those of us in tech to discuss "how are you preparing for / migrating to the AI future"
 

I can't watch this yet cause of work, but will this be available to subscribers? The paid ChatGPT is part of my daily workflow, and I use it a decent bit in my personal life too, so I'm excited for this. (Edit: Nvm, just watched it)

Just last night, I had it do a deep dive to find a service menu issue with an old hotel TV from 2011 after I couldn't find the answer myself, and it actually figured it out. It's getting to the point where it's sometimes better than me at researching and googling obscure things, and it does it so much faster.
 
Last edited:
So it seems like reception is pretty lukewarm to GPT5? Am I missing any major leaps?

Sam Altman was hyping GPT5 like crazy over the past couple weeks. I guess it was all for the $500B valuation round which closed before the GPT5 livestream, lmao. No refunds!
 
Each day I pray that the Eye of Sauron that is AI development does not turn to my job function.

It's only a matter of time until I'm automated out. Luckily they seem to be focusing on entry-level positions and traditional programming for now.

First they came for the content writers, and I did not speak out—
Because I was not a content writer.

Then they came for the graphic designers, and I did not speak out—
Because I was not a graphic designer.

Then they came for the customer support teams, and I did not speak out—
Because I wrote code, not scripts.

Then they came for the junior developers, and I did not speak out—
Because I was a senior engineer at Microsoft.

Then they came for me—
And the systems I helped build were the ones replacing me.
 
Tech isn't even the primary danger zone, it's white collar jobs.

I can tell from personal experience that normies have no idea what's coming.
At the moment in my corner of tech (disclosure though: I've been involved with NLP / language models in various projects off and on since even before the GPT/tranformer revolution, though I do a lot of other tech too) it's just more work to be done, if anything; the total change in what's possible means all kinds of refactors and new AI-integrated systems, and of course it also means learning to rapidly use AI during development itself.
 
Each day I pray that the Eye of Sauron that is AI development does not turn to my job function.

It's only a matter of time until I'm automated out. Luckily they seem to be focusing on entry-level positions and traditional programming for now.
Automation has never led to mass unemployment. My job will be made redundant within 5 years max, it is what it is. No point stressing about it. New jobs will be created even if we can't quite envision what sort of jobs will emerge. Zero sum economic thinking is way too widely spread given that it's only ever produced incorrect predictions.
 
Automation has never led to mass unemployment. My job will be made redundant within 5 years max, it is what it is. No point stressing about it. New jobs will be created even if we can't quite envision what sort of jobs will emerge. Zero sum economic thinking is way too widely spread given that it's only ever produced incorrect predictions.

You're naive if you think AI won't lead to mass unemployment. The number of job functions that could become redundant because of AI is just mind boggling.
 
I'm trying to test the coding in GPT5, but it's not working right now. I'm guessing their servers are getting hammered, but I haven't been able to complete a "5 Thinking" task yet. It made it pretty far in my coding challenge, and then it just gave up. I'm not sure why, it didn't give me a specific error. I wasn't able to see or copy the code that it had already written either, which was disappointing. Hopefully it starts working again soon.
 
Benchmarks are solid, but nothing really mind-blowing. But honestly I stopped paying much attention to benchmarks quite some time ago, I just use the models myself and evaluate accordingly. Drastic reduction in hallucination is a big deal and very welcome. Immediate rollout to the free tier is a pretty good deal, too bad I renewed my Plus annual literally three days ago.

Anyway, this was a pretty weird presentation. No real bangers, quite a lot of filler (why?) and I don't really feel the OpenAI "San Francisco youngster geek squad" vibe, and the cancer lady was weird and unnecessary. Sam Altman is just a charismatic void, a black hole personality. Also nothing on agent systems?

I dunno, I expected a banger but maybe we've reached a point of diminishing returns with these large language models in their current iteration.

99% of normies are not going to notice a difference between 4o and 5.

Edit: and they're getting clowned on for their bizarre graphs.







Did they use AI to make this chart?


jbzhp5h.jpeg
 
Last edited:
I have to say, it's quite a nice model to work with for brainstorming a technical project. I'm currently thinking through a somewhat complex stack that is exceedingly difficult even to describe to a person, and am very deep in the conversation now with GPT5, and this time I can say that the insights on very tricky subtleties are much better than I got earlier on Claude on the same topic.
 
I can't watch this yet cause of work, but will this be available to subscribers? The paid ChatGPT is part of my daily workflow, and I use it a decent bit in my personal life too, so I'm excited for this. (Edit: Nvm, just watched it)

Just last night, I had it do a deep dive to find a service menu issue with an old hotel TV from 2011 after I couldn't find the answer myself, and it actually figured it out. It's getting to the point where it's sometimes better than me at researching and googling obscure things, and it does it so much faster.
Free users, plus users, and pro users all get access to GPT-5 w/ various settings and limits. Plus users get nearly unlimited for normal use.
 
You're naive if you think AI won't lead to mass unemployment. The number of job functions that could become redundant because of AI is just mind boggling.

You're naive if you think that a prediction (new innovation will lead to mass unemployment) that has been uttered five million times during the history of mankind and proven wrong every single time is finally coming true. Even the "this time it's different!" claim is old. Yes, we have a good idea of which jobs will be made redundant but we don't have the capacity to understand which jobs will be created. It's just lump of labor fallacy #6353453.

Is this in reference to the last 150 years, or are we talking about the last 20 years?

KHS5eaV.png


This is about 80 years worth of data. The States are more automated than ever in 2025 yet unemployment is very low. How come? What is the correlation between automation and the unemployment rate in your opinion?
 
Last edited:
KHS5eaV.png


This is about 80 years worth of data. The States are more automated than ever in 2025 yet unemployment is very low. How come? What is the correlation between automation and the unemployment rate in your opinion?
Great questions, thank you for that data.

Edit: I see many folks that want to step up and do the work that is required. These days we must. It was way different 120 years ago, like losing your job to a machine was so visceral.
 
Last edited:
I have paid api verified organisation with open ai.

I configured roocode vscode extension for gpt5 and asked it to code a threejs rubiks cube. It coded away for 10 mins and then said it hit max tokens.. so failed.
 
Top Bottom