VQAv2 (test)

VQA v2.0 (Visual Question Answering v2.0) is a balanced dataset designed to counter language priors in visual question answering. It consists of complementary image pairs where the same question yields different answers, forcing models to rely on visual understanding rather than language bias. The dataset contains 1,105,904 questions across 204,721 COCO images, requiring understanding of vision, language, and commonsense knowledge.

Llama 3.2 11B Instruct from Meta currently leads the VQAv2 (test) leaderboard with a score of 0.752 across 1 evaluated AI models.

Paper

MetaLlama 3.2 11B Instruct leads with 75.2%.

Progress Over Time

Interactive timeline showing model performance evolution on VQAv2 (test)

State-of-the-art frontier
Open
Proprietary

VQAv2 (test) Leaderboard

1 models
ContextCostLicense
111B128K$0.05 / $0.05
Notice missing or incorrect data?

FAQ

Common questions about VQAv2 (test).

What is the VQAv2 (test) benchmark?

VQA v2.0 (Visual Question Answering v2.0) is a balanced dataset designed to counter language priors in visual question answering. It consists of complementary image pairs where the same question yields different answers, forcing models to rely on visual understanding rather than language bias. The dataset contains 1,105,904 questions across 204,721 COCO images, requiring understanding of vision, language, and commonsense knowledge.

What is the VQAv2 (test) leaderboard?

The VQAv2 (test) leaderboard ranks 1 AI models based on their performance on this benchmark. Currently, Llama 3.2 11B Instruct by Meta leads with a score of 0.752. The average score across all models is 0.752.

What is the highest VQAv2 (test) score?

The highest VQAv2 (test) score is 0.752, achieved by Llama 3.2 11B Instruct from Meta.

How many models are evaluated on VQAv2 (test)?

1 models have been evaluated on the VQAv2 (test) benchmark, with 0 verified results and 1 self-reported results.

Where can I find the VQAv2 (test) paper?

The VQAv2 (test) paper is available at https://arxiv.org/abs/1612.00837. The paper details the methodology, dataset construction, and evaluation criteria.

What categories does VQAv2 (test) cover?

VQAv2 (test) is categorized under image to text, multimodal, reasoning, and vision. The benchmark evaluates multimodal models.

More evaluations to explore

Related benchmarks in the same category

View all image to text
GPQA

A challenging dataset of 448 multiple-choice questions written by domain experts in biology, physics, and chemistry. Questions are Google-proof and extremely difficult, with PhD experts reaching 65% accuracy.

reasoning
213 models
MMLU-Pro

A more robust and challenging multi-task language understanding benchmark that extends MMLU by expanding multiple-choice options from 4 to 10, eliminating trivial questions, and focusing on reasoning-intensive tasks. Features over 12,000 curated questions across 14 domains and causes a 16-33% accuracy drop compared to original MMLU.

reasoning
119 models
AIME 2025

All 30 problems from the 2025 American Invitational Mathematics Examination (AIME I and AIME II), testing olympiad-level mathematical reasoning with integer answers from 000-999. Used as an AI benchmark to evaluate large language models' ability to solve complex mathematical problems requiring multi-step logical deductions and structured symbolic reasoning.

reasoning
107 models
MMLU

Massive Multitask Language Understanding benchmark testing knowledge across 57 diverse subjects including STEM, humanities, social sciences, and professional domains

reasoning
99 models
SWE-Bench Verified

A verified subset of 500 software engineering problems from real GitHub issues, validated by human annotators for evaluating language models' ability to resolve real-world coding issues by generating patches for Python codebases.

reasoning
89 models
Humanity's Last Exam

Humanity's Last Exam (HLE) is a multi-modal academic benchmark with 2,500 questions across mathematics, humanities, and natural sciences, designed to test LLM capabilities at the frontier of human knowledge with unambiguous, verifiable solutions

reasoningmultimodal
74 models