math benchmark
We-Math
We-Math evaluates multimodal models on visual mathematical reasoning, requiring models to understand and solve math problems presented with visual elements such as diagrams, charts, and geometric figures.
Updated Aug 7, 2026
Published models1
Registry coverage1
MetricScore
EvidenceB
We-Math leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | AC | 89.0% | 100.0% | 1 | C |
We-Math highlights
The top published results on this benchmark's own scale.
What is We-Math?
Definition and scoring fields from the benchmark registry.
We-Math evaluates multimodal models on visual mathematical reasoning, requiring models to understand and solve math problems presented with visual elements such as diagrams, charts, and geometric figures.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- We-Math
- Modality
- multimodal
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- we-math|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about We-Math.
Which model scores highest on We-Math?
Qwen3.6 Plus is currently ranked first with 89.0%.
What does We-Math measure?
We-Math evaluates multimodal models on visual mathematical reasoning, requiring models to understand and solve math problems presented with visual elements such as diagrams, charts, and geometric figures.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
1 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.