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

Image Understanding research benchmarks – Page 2 · SOTA2 Research

Image Understanding

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

MMBench v1.1 (test)

85MMB^i Score

10

Feb 26, 2026

AI2D 2016 (test)

89.2Score

9

Jun 12, 2026

Image Understanding Suite GQA POPE SQA MME MMB (test val)

60.5GQA

9

Jun 2, 2026

68.1Accuracy (CLS)

9

Mar 4, 2026

ChartQA 2022 (test)

86.5Overall Score

8

Jun 12, 2026

LLaVABenchWilder

LLaVA-OneVision-7B w/ VL-MDR

77.9Score

8

Apr 8, 2026

LLaVA-OneVision-7B w/ Skywork-VL

101.9Score

8

Apr 8, 2026

40.2MMIU Score

7

Feb 26, 2026

67.3SEED-2 Image Score

6

Feb 26, 2026

STCray and MMXray (test)

12.3BLEU-4

5

Jun 16, 2026

7 Image Tasks (VQAv2, GQA, TextVQA, VizWiz, ScienceQA, POPE, OK-VQA)

62.93Accuracy

5

May 20, 2026

InfoVQA 2022 (val)

78.6Score

4

Jun 12, 2026

DocVQA 2021 (val)

93Score

4

Jun 12, 2026

Jetson Orin Nano Performance Benchmark

88Vision Encoding Time (ms)

4

Feb 26, 2026

56Vision Encoding Time (ms)

4

Feb 26, 2026

COCO GPT-based evaluation

84.1Conversation Score

4

Feb 26, 2026

37.9Accuracy

2

Feb 26, 2026

75.7Accuracy

2

Feb 26, 2026

76.4Accuracy

2

Feb 26, 2026

iPhone 17 Pro Performance Benchmark

102Vision Encoding Time (ms)

1

Feb 26, 2026

Page 2 of 2

Popular datasets

MMBench MME TextVQA MMStar COCO

Follow for updates

Join SOTA2 Contact