What it measures
- Visual math
- Diagram reasoning
- Quantitative multimodal QA
MathVista
A visual math reasoning benchmark for multimodal models.
How to read the score
% accuracy; higher is better.
Sourced Ranking
| Rank | Model | Provider | Score | Source | Source Type | Verified |
|---|
Limitations
- Do not compare testmini and test scores directly without labeling.
- The paper is static; the official leaderboard is current.
- Track testmini and test score definitions separately.
leaderboard
MathVista official leaderboard
The paper is static; the official leaderboard is current. Track testmini and test score definitions separately.
paper
MathVista paper or specification
Use for benchmark definition, not necessarily for latest model scores.
official
MathVista dataset or code
Use for dataset, code, or implementation details; score freshness depends on the benchmark.
leaderboard
Kaggle MathVista benchmark page
Kaggle benchmark page can be used as an operational leaderboard source. Preserve split labels such as testmini and test when adding scores.