Last released Sep 7, 2026
Evaluate models, agents and harnesses: program checks first, rubrics for the rest, honest paired statistics.