Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

Open-Ended Visual Question Answering research benchmarks · SOTA2 Research

Open-Ended Visual Question Answering

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

VQA 1.0 (test-dev)

Ensemble of 7 Att. models

66.7Overall Accuracy

100

Feb 26, 2026

VQA 1.0 (test-standard)

67.36Overall Accuracy

50

Feb 27, 2026

VQA (test-standard)

83.3Accuracy (Overall)

32

Feb 26, 2026

0.525BLEU-1

26

Feb 26, 2026

36.8Accuracy

23

Feb 26, 2026

PMC-VQA (test-initial)

35.4BLEU-1

19

Feb 26, 2026

45.32Accuracy

14

Jun 25, 2026

49.43Accuracy

14

Jun 25, 2026

64.92Accuracy

14

Jun 25, 2026

31.73Accuracy

14

Jun 25, 2026

32.74H-CQS METEOR

13

Apr 16, 2026

EarthVLSet 1.0 (test)

0.573BLEU-1

12

Feb 26, 2026

LLaVA-in-the-Wild (LLaVA-W)

68.5Score

10

May 18, 2026

0.86BERTScore

8

Feb 26, 2026

LLaVA Bench v1 (test)

37.18Relevance

7

Feb 26, 2026

LLaVA Eval v1 (test)

77.67Conversation Score

7

Feb 26, 2026

56.7F1 Score

6

Mar 10, 2026

67.8F1 Score

6

Mar 10, 2026

63.1F1 Score

6

Mar 10, 2026

72.91F1 Score

6

Mar 10, 2026

Llama-2 Chat 7B

56.39Accuracy

6

Feb 26, 2026

53.8WSI-P (Morphology)

5

Mar 4, 2026

Popular datasets

VQA PMC-VQA Vizwiz LLaVA-Bench SLAKE

Follow for updates

Join SOTA2 Contact