Benchmark
BeyondAIME
An AI evaluation benchmark targeting high-level mathematics competition problems. It tests a model's multi-step reasoning and logical rigor on problems at the AIME difficulty level.
Timeline (last 90 days)
Sep 12
The SAO team selected three high-difficulty benchmarks—SWE-Bench Verified, BeyondAIME, and IMOAnswerBench—for validation
Unverified50%
Sep 12
BeyondAIME and IMOAnswerBench target high-level math competition problems at the AIME and IMO levels respectively
Unverified50%