Re:GPT-5
by daveguy on Thursday August 07, 2025 @02:28PM
(#44828491)
I would love to see how this performs on ARC-AGI 2, zero-shot, private eval. I hope we get an update from Chollet and team regarding performance.
[ Parent
| on HN ]
9.9% on ARC-AGI 2
Hah, that was fast! Thank you. They must have had preview access. It didn't bode well that SimonW [0] had to explicitly tell GPT-5 to use python to get a table sorted correctly (but awesome that in can use python as a tool without any plumbing). It appears we are not quite to AGI yet.