Report
Opus 5.5 reportedly reaches a new high score on ProgramBench
One poster says every model scored 0% at launch, prompting weaker ProgramBench metrics such as “almost resolved.”
TLDR
A ProgramBench poster says every model scored 0% when the benchmark launched, prompting weaker metrics such as “almost resolved.” The poster says things have since changed. In a separate post, another user congratulates Anthropic’s team on a new Opus 5.5 high score on ProgramBench.
Combined views
494
3 Sources, first seen ago
