News for hackers. Stuff that matters.

A comment on Claude Opus 4.5. Parent comment

Re:Claude Opus 4.5 by imiric on Monday November 24, 2025 @04:07PM (#46039332)
Or, 2b: the nerf is real, but benchmarks are gamed and models are trained to excel at them, yet fall flat in real world situations.
[ Parent | on HN ]
Re:Claude Opus 4.5 by metalliqaz on Monday November 24, 2025 @04:14PM (#46039416)
I mostly stay out of the LLM space but I thought it was an open secret already that the benchmarks are absolutely gamed.
[ Parent | on HN ]