multimodal benchmark
InfoVQA
InfoVQA dataset with 30,000 questions and 5,000 infographic images requiring joint reasoning over document layout, textual content, graphical elements, and data visualizations with elementary reasoning and arithmetic skills
Updated Aug 7, 2026
Published models9
Registry coverage9
MetricScore
EvidenceB
InfoVQA leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | AC | 83.4% | 100.0% | 9 | C | |
| 2 | AC | 82.6% | 87.5% | 9 | C | |
| 3 | DE | 78.1% | 75.0% | 9 | C | |
| 4 | DE | 75.8% | 62.5% | 9 | C | |
| 5 | MI | 72.7% | 50.0% | 9 | C | |
| 6 | GO | 70.6% | 37.5% | 9 | C | |
| 7 | DE | 66.1% | 25.0% | 9 | C | |
| 8 | GO | 64.9% | 12.5% | 9 | C | |
| 9 | GO | 50.0% | 0.0% | 9 | C |
Score distribution
Top published rows on the benchmark's original scale.
InfoVQA highlights
The top published results on this benchmark's own scale.
What is InfoVQA?
Definition and scoring fields from the benchmark registry.
InfoVQA dataset with 30,000 questions and 5,000 infographic images requiring joint reasoning over document layout, textual content, graphical elements, and data visualizations with elementary reasoning and arithmetic skills
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- InfoVQA
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- infovqa|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about InfoVQA.
Which model scores highest on InfoVQA?
Qwen2.5 VL 32B Instruct is currently ranked first with 83.4%.
What does InfoVQA measure?
InfoVQA dataset with 30,000 questions and 5,000 infographic images requiring joint reasoning over document layout, textual content, graphical elements, and data visualizations with elementary reasoning and arithmetic skills
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
9 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.