math benchmark
MATH
MATH dataset contains 12,500 challenging competition mathematics problems from AMC 10, AMC 12, AIME, and other mathematics competitions. Each problem includes full step-by-step solutions and spans multiple difficulty levels (1-5) across seven mathematical subjects including Prealgebra, Algebra, Number Theory, Counting and Probability, Geometry, Intermediate Algebra, and Precalculus.
Updated Aug 11, 2026
Models71
Model coverage71
MetricScore
EvidenceB
MATH Ranking
Higher score ranks better on this benchmark.
MATH Score Distribution
A closer view of the leading scores on this benchmark.
MATH Highlights
The leading models and scores on this benchmark.
What is MATH?
What MATH measures and how its scores work.
MATH dataset contains 12,500 challenging competition mathematics problems from AMC 10, AMC 12, AIME, and other mathematics competitions. Each problem includes full step-by-step solutions and spans multiple difficulty levels (1-5) across seven mathematical subjects including Prealgebra, Algebra, Number Theory, Counting and Probability, Geometry, Intermediate Algebra, and Precalculus.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- MATH
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- math|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about MATH.
Which model scores highest on MATH?
o3-mini is currently ranked first with 97.9%.
What does MATH measure?
MATH dataset contains 12,500 challenging competition mathematics problems from AMC 10, AMC 12, AIME, and other mathematics competitions. Each problem includes full step-by-step solutions and spans multiple difficulty levels (1-5) across seven mathematical subjects including Prealgebra, Algebra, Number Theory, Counting and Probability, Geometry, Intermediate Algebra, and Precalculus.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
71 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.