Loading…
Radar · Entity
The newsletter reports Claude Opus 5.5 scoring 18.5% resolved and 70.5% almost on ProgramBench, compared with 0% resolved for the best system when the benchmark appeared.
No tracked model releases for this entity yet.
No linked research papers yet.
1 item on this desk · stories, not raw articles, newest first