math benchmark
MathVision
MATH-Vision is a dataset designed to measure multimodal mathematical reasoning capabilities. It focuses on evaluating how well models can solve mathematical problems that require both visual understanding and mathematical reasoning, bridging the gap between visual and mathematical domains.
Updated Aug 7, 2026
Published models32
Registry coverage32
MetricScore
EvidenceB
MathVision leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
Score distribution
Top published rows on the benchmark's original scale.
MathVision highlights
The top published results on this benchmark's own scale.
What is MathVision?
Definition and scoring fields from the benchmark registry.
MATH-Vision is a dataset designed to measure multimodal mathematical reasoning capabilities. It focuses on evaluating how well models can solve mathematical problems that require both visual understanding and mathematical reasoning, bridging the gap between visual and mathematical domains.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- MathVision
- Modality
- multimodal
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- mathvision|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about MathVision.
Which model scores highest on MathVision?
Kimi K3 is currently ranked first with 97.8%.
What does MathVision measure?
MATH-Vision is a dataset designed to measure multimodal mathematical reasoning capabilities. It focuses on evaluating how well models can solve mathematical problems that require both visual understanding and mathematical reasoning, bridging the gap between visual and mathematical domains.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
32 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is preserved as source-native evidence but is not eligible for the current overall score.