ArXivMath
matharena.ai- Category
- Developer Tools
- Type
- TOOL
- Date
About
Research mathematics questions drawn from recent arXiv papers. The August 2026 edition contains 57 questions, prioritizes resolved conjectures, and tests models in native agent harnesses with Python and SageMath but no internet. Final answers are checked by an LLM judge. Dataset editions and harness settings must be compared separately.
Why it made the leaderboard
Compare dataset edition, agent harness, tool access and grading before comparing results. A single-run observation does not establish a cross-model ranking.
Intel on ArXivMath
Tags
benchmarkevaluation
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.