News for hackers. Stuff that matters.

A comment on MaxxBench – "Nice DeepSWE score. But is it true?". Parent comment

Re:MaxxBench – "Nice DeepSWE score. But is it true?" by nullc on Saturday November 07, 2026 @05:21PM (#50408608)
5.3 introduced tons of claudish babble, really a massive step down in its communication style.
[ Parent | on HN ]
Re:MaxxBench – "Nice DeepSWE score. But is it true?" by enraged_camel on Saturday November 07, 2026 @06:14PM (#50408956)
It's because they distilled Opus 5.
[ Parent | on HN ]
Re:MaxxBench – "Nice DeepSWE score. But is it true?" by nullc on Sunday November 08, 2026 @12:39PM (#50415339)
Or because they want it to not be obvious when they're rerouting requests to Opus for the purpose of harvesting traces from it. :(
[ Parent | on HN ]