II.
Benchmark overview
Reference · livebenchmark:olympiad-bench
OlympiadBench overview
OlympiadBench (OpenBMB) is a bilingual (English/Chinese) multimodal benchmark of 8,476 olympiad-level math and physics problems, with expert-annotated step-by-step solutions, used to probe the limits of frontier-model reasoning.
Attributes
displayName
OlympiadBench
homepageUrl
kind
model-only
targetsKind
ModelVersion
description
OlympiadBench (OpenBMB) is a bilingual (English/Chinese) multimodal
benchmark of 8,476 olympiad-level math and physics problems, with
expert-annotated step-by-step solutions, used to probe the limits
of frontier-model reasoning.
Outgoing edges
applies_to2
- domain:mathematics·DomainMathematics
- domain:physics·DomainPhysics
covers1
- skill-area:mathematical-reasoning·SkillAreaMathematical Reasoning
Incoming edges
None.