Werewolf wrote: ↑Sat Jul 26, 2025 7:54 pmWait 2 weeks...ChatGPT-5 should be out.
Better than 50% chance it will be out by 15-Aug-25 - link.
You have to pay to use Grok 4: I suspect the same will be true of GPT5.
Markets are confident it will be out in the next 5 days - link.
Will be interesting to see how good it is, and if it is the best, how long it will take competitors to catch it.
Yes I’m waiting in anticipation. I don’t think the competition will catch it soon, because current ChatGPT (4O + O3) is about equal to latest Grok / Gemini / Claude overall, and ChatGPT-5 is apparently a big step up.
First thing I want to do is write a chess engine with it.
Grok 4 was summarily defeated by o3 in a best-of-three match. Carlsen and Howell provided commentary on the duel. Grok 4 had previously defeated Gemini 2.5 Pro by a score of 3 to 2, while o3 beat o4 mini 4–0 in the semifinals.
Werewolf wrote: ↑Sat Jul 26, 2025 7:54 pmWait 2 weeks...ChatGPT-5 should be out.
Better than 50% chance it will be out by 15-Aug-25 - link.
You have to pay to use Grok 4: I suspect the same will be true of GPT5.
Markets are confident it will be out in the next 5 days - link.
Will be interesting to see how good it is, and if it is the best, how long it will take competitors to catch it.
Yes I’m waiting in anticipation. I don’t think the competition will catch it soon, because current ChatGPT (4O + O3) is about equal to latest Grok / Gemini / Claude overall, and ChatGPT-5 is apparently a big step up.
First thing I want to do is write a chess engine with it.
It looks as though the release of GPT 5 is in process: it has gone straight to the top of the leader board - link.
Human chess is partly about tactics and strategy, but mostly about memory
I am very unconvinced by ChatGPT 5's reasoning ability.
We played 6 games of chess - it made illegal moves in 5 of them. All my attempts to get it to think (invoke inference) failed. It seems like OpenAI are trying to save money by taking inference out of the hands of the user.
Then I tested it on a 1990s IQ puzzle book that's not on the net. It scored a dismal 50%.
Grok 4 got 95%.
Werewolf wrote: ↑Fri Aug 08, 2025 9:47 am
I am very unconvinced by ChatGPT 5's reasoning ability.
We played 6 games of chess - it made illegal moves in 5 of them. All my attempts to get it to think (invoke inference) failed. It seems like OpenAI are trying to save money by taking inference out of the hands of the user.
Then I tested it on a 1990s IQ puzzle book that's not on the net. It scored a dismal 50%.
Grok 4 got 95%.
ChatGPT-O3 got 70%.
We are still in a world in which different chatbots are suited to different tasks. Another example: GPT-5 has an input window of 272,000 tokens, whereas Gemini has a million.
Open AI have claimed that GPT-5 is better at medical diagnosis than the other chatbots: I have no reason to disbelieve this claim. If you suspect you have a mental health issue, go to Grok - it will actually give you an answer, while other chatbots will refuse to do so.
And so on and so on...
Human chess is partly about tactics and strategy, but mostly about memory
My feelings about ChatGPT-5 remain mixed, however, after 24 hours of work I have a functioning chess engine - my first.
It’s a mainly brute force engine with a primitive evaluation function written in C. I decided to make Threads dynamic so that they equal the number of legal moves available (the idea was to do multi threading at the root).
It works within Fritz and plays at around 1800 FIDE Elo.
Werewolf wrote: ↑Sat Aug 09, 2025 6:31 pmMy feelings about ChatGPT-5 remain mixed, however, after 24 hours of work I have a functioning chess engine - my first.
Good work - well done!
You might have to do it again in a few weeks - the market is expecting GPT-5 to lose the throne very soon - link.
Human chess is partly about tactics and strategy, but mostly about memory
Werewolf wrote: ↑Sat Aug 09, 2025 6:31 pmMy feelings about ChatGPT-5 remain mixed, however, after 24 hours of work I have a functioning chess engine - my first.
Good work - well done!
You might have to do it again in a few weeks - the market is expecting GPT-5 to lose the throne very soon - link.
Gemini 3?
Writing a chess engine is less fun than I imagined, I am literally spending 95% of my time de-bugging
