arXiv cs.AI· Jingjing Wei, Hao Zhao, Xinke Li, Linxi Yu, Xiaosong Gao, Shanshan Xiao·· 6 天前AI 评分30
MathVis：AI 能否通过画图推理？数学图形生成基准发布
MathVis: Can AI Reason through Drawing? A Benchmark for Mathematical Figure Generation
AI 导读
研究者发布 MathVis 基准，用 512 个任务评估 LLM 可视化前沿数学问题并进行图示辅助推理的能力，每个任务包含数学陈述、参考图、可执行 TikZ 或 SVG 代码、Gold IR、原子评分细则和错误码。
正文 · 原文
Abstract:This paper introduces MathVis, a benchmark designed to evaluate LLMs' ability to visualize frontier mathematical problems and perform diagram-assisted reasoning. The benchmark consists of 512 tasks, each comprising a mathematical statement, a reference diagram, executable TikZ or SVG code, a Gold Intermediate Representation (IR), atomic rubrics, and error codes. Through rigorous evaluation, GPT-6-Astra, Claude Fable 5, and Gemini 3.1 Pro Preview achieve pass@1 scores of 31.25%, 26.56%, and 22.85%, respectively. We found that although the tasks require visualization, LLMs produce outputs dominated by formal languages, symbols, and graphics programs, rather than explicit diagram-mediated reasoning. Furthermore, while the solutions generated by LLMs may contain all the necessary elements, they are often visually confusing to humans. Ultimately, this benchmark aims to visualize mathematical problems and promote research on enabling LLMs to simplify their reasoning via visual diagrams.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.12954 [cs.AI] |
| (or arXiv:2610.12954v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.12954 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jingjing Wei [view email]
[v1]
Fri, 9 Oct 2026 03:52:59 UTC (6,277 KB)
来源：arXiv cs.AI · arxiv.org