News for hackers. Stuff that matters.

A comment on Claude Opus 5. Parent comment

Re:Claude Opus 5 by wyre on Friday July 24, 2026 @01:48PM (#49039265)
If they were benchmaxxing, surely they would score higher than 30% on ARC-AGI.
[ Parent | on HN ]
Re:Claude Opus 5 by conradkay on Friday July 24, 2026 @02:29PM (#49039764)
Doing a quick search it seems like the average human score is 49%?

I view benchmaxxing as more of a spectrum. Mmaybe they're doing a lot more RL in environments similar to ARC-AGI 3, not even with the purpose of scoring well on any benchmark but hoping it generalizes into better performance on real, useful tasks.

[ Parent | on HN ]