math benchmark
ArXivMath
ArXivMath is a final-answer benchmark of research-level mathematics maintained by MathArena. Problems are extracted monthly from recent arXiv paper abstracts, then filtered through automated and manual checks to ensure they are self-contained, non-trivial, and verifiable. Because problems are drawn from active research, the benchmark is more realistic and more closely connected to mathematical research than contest or olympiad benchmarks.
Updated Aug 11, 2026
Models2
Model coverage2
MetricScore
EvidenceB
ArXivMath Ranking
Higher score ranks better on this benchmark.
| 01 | AN | 72.2% | 100.0% | 2 | C | |
| 02 | TE | 52.2% | 0.0% | 2 | C |
ArXivMath Score Distribution
A closer view of the leading scores on this benchmark.
ArXivMath Highlights
The leading models and scores on this benchmark.
What is ArXivMath?
What ArXivMath measures and how its scores work.
ArXivMath is a final-answer benchmark of research-level mathematics maintained by MathArena. Problems are extracted monthly from recent arXiv paper abstracts, then filtered through automated and manual checks to ensure they are self-contained, non-trivial, and verifiable. Because problems are drawn from active research, the benchmark is more realistic and more closely connected to mathematical research than contest or olympiad benchmarks.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- ArXivMath
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- arxivmath|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about ArXivMath.
Which model scores highest on ArXivMath?
Claude Sonnet 5 is currently ranked first with 72.2%.
What does ArXivMath measure?
ArXivMath is a final-answer benchmark of research-level mathematics maintained by MathArena. Problems are extracted monthly from recent arXiv paper abstracts, then filtered through automated and manual checks to ensure they are self-contained, non-trivial, and verifiable. Because problems are drawn from active research, the benchmark is more realistic and more closely connected to mathematical research than contest or olympiad benchmarks.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
2 model results are currently shown.
Does this benchmark affect the overall score?
No. This benchmark is shown for reference but does not contribute to the overall score.