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.

o3-mini from OpenAI currently leads the MATH leaderboard with a score of 0.979 across 70 evaluated AI models.

Paper

OpenAIo3-mini leads with 97.9%, followed by OpenAIo1 at 96.4% and Mistral AIMistral Large 3 at 90.4%.

Progress Over Time

Interactive timeline showing model performance evolution on MATH

State-of-the-art frontier
Open
Proprietary

MATH Leaderboard

70 models
ContextCostLicense
1
OpenAI
OpenAI
2
OpenAI
OpenAI
3
Mistral AI
Mistral AI
675B
314B
5
6
Moonshot AI
Moonshot AI
1.0T
727B
88B
9
10
11
12
OpenAI
OpenAI
1312B
14
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
33B
14
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
73B
163B
17
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
34B
18
Microsoft
Microsoft
15B
19
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
15B
20
21
2270B
23
Amazon
Amazon
23
OpenAI
OpenAI
128K$2.50 / $10.00
25
264B
27
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
8B
28236B
29405B
30
Amazon
Amazon
31
32128K$10.00 / $30.00
33
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
235B
34
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
7B
35
3624B
37
Moonshot AI
Moonshot AI
1.0T
37
3924B
40
4124B
41
4390B
44
Microsoft
Microsoft
4B
45400B
46
Anthropic
Anthropic
47
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
72B
4860B
498B
50
Alibaba Cloud / Qwen Team
Alibaba Cloud / Qwen Team
32B
150 of 70
1/2
Notice missing or incorrect data?

FAQ

Common questions about MATH.

What is the MATH benchmark?

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.

What is the MATH leaderboard?

The MATH leaderboard ranks 70 AI models based on their performance on this benchmark. Currently, o3-mini by OpenAI leads with a score of 0.979. The average score across all models is 0.668.

What is the highest MATH score?

The highest MATH score is 0.979, achieved by o3-mini from OpenAI.

How many models are evaluated on MATH?

70 models have been evaluated on the MATH benchmark, with 0 verified results and 68 self-reported results.

Where can I find the MATH paper?

The MATH paper is available at https://arxiv.org/abs/2103.03874. The paper details the methodology, dataset construction, and evaluation criteria.

What categories does MATH cover?

MATH is categorized under math and reasoning. The benchmark evaluates text models.

More evaluations to explore

Related benchmarks in the same category

View all math
GPQA

A challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. Questions are Google-proof and extremely difficult, with PhD experts reaching 65% accuracy.

reasoning
214 models
MMLU-Pro

A more robust and challenging multi-task language understanding benchmark that extends MMLU by expanding multiple-choice options from 4 to 10, eliminating trivial questions, and focusing on reasoning-intensive tasks. Features over 12,000 curated questions across 14 domains and causes a 16-33% accuracy drop compared to original MMLU.

math
119 models
AIME 2025

All 30 problems from the 2025 American Invitational Mathematics Examination (AIME I and AIME II), testing olympiad-level mathematical reasoning with integer answers from 000-999. Used as an AI benchmark to evaluate large language models' ability to solve complex mathematical problems requiring multi-step logical deductions and structured symbolic reasoning.

math
108 models
MMLU

Massive Multitask Language Understanding benchmark testing knowledge across 57 diverse subjects including STEM, humanities, social sciences, and professional domains

math
99 models
SWE-Bench Verified

A verified subset of 500 software engineering problems from real GitHub issues, validated by human annotators for evaluating language models' ability to resolve real-world coding issues by generating patches for Python codebases.

reasoning
89 models
Humanity's Last Exam

Humanity's Last Exam (HLE) is a multi-modal academic benchmark with 2,500 questions across mathematics, humanities, and natural sciences, designed to test LLM capabilities at the frontier of human knowledge with unambiguous, verifiable solutions

mathmultimodal
74 models