News for hackers. Stuff that matters.

A comment on Senior SWE-Bench: open-source benchmark that assesses agents as senior engineers. Parent comment

Re:Senior SWE-Bench: open-source benchmark that assesses age... by pishpash on Thursday July 02, 2026 @01:09AM (#48756789)
Maybe models should ask for human-in-the-loop input, as a matter of convention.
[ Parent | on HN ]
Re:Senior SWE-Bench: open-source benchmark that assesses age... by sinuhe69 on Thursday July 02, 2026 @02:24AM (#48757306)
A model that can ask questions or ask for help when in doubt is indeed a major feat. None of the current frontier models can do that.
[ Parent | on HN ]