PhysClosureBench

Search everything

Mentions · 7d0
Mentions · 30d1
All mentions1
First seenOct 14, 2026
Last mentioned16d ago

Timeline

1 articles · newest first

PhysClosureBench finds frontier agents often miss full physics research closure

Claude-Opus-5 scores 6.29/10 on average but reaches full closure in only 32.7% of attempts.

Hugging Face daily papersOct 14, 20:00 UTC · 16d agoAI research

Appears with

Claude-Opus-5ai model · 1

Entities are extracted by the model from each article. First seen Oct 14, 2026, last mentioned 16d ago. Watching keeps it in this browser only (no account); the watchlist page and dashboard alerts use it.