ResearchBenchmarksReasoning and Question Answering on HLE (reference)Follow185Problems EvaluatedTheoria175.75180.375185189.625Jul 1, 2026Evaluation ResultsMethodMethodLinksProblems EvaluatedCoverage CountCertified PrecisionWilson 95% CI Lower BoundSolver Raw AccuracyTheoria2026.071851059684.5155