Re:GPT-5.5
by nullbyte on Thursday April 23, 2026 @02:09PM
(#47879206)
82.7% on Terminal Bench is crazy
[ Parent
| on HN ]
Is it? There are 5 other models near ~80% and it was achieved in March... which in AI-world seems like a century ago.
those are not verified. I've tried forgecode and I cannot believe they didn't do something to influence the benchmarks
Re:GPT-5.5
by GodelNumbering on Thursday April 23, 2026 @05:07PM
(#47882012)
Yup, they were found to be sneaking the answer key using agents.md
https://debugml.github.io/cheating-agents/#sneaking-the-answ...