News for hackers. Stuff that matters.
Re:MaxxBench – "Nice DeepSWE score. But is it true?" by dgacmu on Sunday November 08, 2026 @06:25PM (#50418780)
No, this modified the original prompt, it didn't do a multi-round evaluation like slop code. It was a little hard to get that from the writeup, I had the same initial reaction.

There are no comments on this comment yet.