image to text benchmark
VQAv2 Leaderboard
VQAv2 is a balanced Visual Question Answering dataset with complementary images for each question and approximately twice as many image-question pairs as the original VQA dataset.
Updated Sep 5, 2026
Models3
Model coverage3
MetricScore
EvidenceB
VQAv2 Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelMA | Score80.90% | Percentile100.00% | Participants3 | EvidenceC | Evaluated |
| Rank02 | ModelMA | Score78.60% | Percentile50.00% | Participants3 | EvidenceC | Evaluated |
| Rank03 | ModelME | Score78.10% | Percentile0.00% | Participants3 | EvidenceC | Evaluated |
VQAv2 Highlights
The leading models and scores on this benchmark.
VQAv2 Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for VQAv2
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is VQAv2?
What VQAv2 measures and how its scores work.
VQAv2 is a Visual Question Answering dataset in the image_to_text category.
It reports a Score measured as a ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- VQAv2
- Modality
- multimodal
- Primary category
- image to text
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about VQAv2.
Which model scores highest on VQAv2?
Pixtral Large is currently ranked first with 80.90%.
What are the top three models on VQAv2?
The current leaders are Pixtral Large (80.90%), Pixtral-12B (78.60%), and Llama 3.2 90B Instruct (78.10%).
Which VQAv2 model has the lowest official input price?
Pixtral-12B has the lowest matched official input price at $0.15 input / $0.15 output per 1M tokens.
Which models are fastest among VQAv2 results?
The fastest matched records are Llama 3.2 90B Instruct (24.00 tok/s via DeepInfra), Pixtral Large (0.10 tok/s via Mistral AI), and Pixtral-12B (0.10 tok/s via Mistral AI).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.