II.
Benchmark overview
Reference · livebenchmark:frontier-math
FrontierMath overview
Epoch AI benchmark of original, expert-vetted research-level math problems requiring deep mathematical reasoning, designed to be hard for frontier models even with tool use.
Attributes
displayName
FrontierMath
homepageUrl
kind
math
targetsKind
ModelVersion
description
Epoch AI benchmark of original, expert-vetted research-level math
problems requiring deep mathematical reasoning, designed to be
hard for frontier models even with tool use.
Outgoing edges
applies_to1
- domain:mathematics·DomainMathematics
covers2
- skill-area:mathematical-reasoning·SkillAreaMathematical Reasoning
- skill-area:closed-book-frontier-reasoning·SkillAreaClosed-Book Frontier Reasoning
Incoming edges
None.