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

Multi-modal Understanding research benchmarks – Page 2 · SOTA2 Research

Multi-modal Understanding

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

MM-Bench-CN (MMBCN) (test)

84MMBCN Score

13

Apr 21, 2026

MM-Bench (MMB) (test)

86.3MMB Score

13

Apr 21, 2026

65.3MMBench Accuracy (En)

12

Feb 26, 2026

64.3Accuracy (77.8% reduction ratio)

11

Mar 10, 2026

67Score

10

May 28, 2026

LLaVA-1.5 Vanilla 13B

61.6Accuracy

10

Mar 4, 2026

MME v1.0 (test)

1,531.3MME^P Score

10

Feb 26, 2026

90.85M(cl, nc)

9

Feb 26, 2026

General Multi-modal Evaluation Suite (VQAv2, GQA, VisWiz, ScienceQA-IMG, TextVQA, POPE, MMBench, MM-Vet) standard (test val)

MoE-LLaVA-Phi DeRS-SM

77.7VQAv2 Accuracy

9

Feb 26, 2026

MMBench EN 1.1 (dev)

84.31Accuracy

8

Feb 26, 2026

62.9Overall Score

8

Feb 26, 2026

CV-Bench, BLINK, RealWorldQA, MathVista, MMStar, MMVet

53.8Average Score

8

Feb 26, 2026

2,315MME Score

7

Apr 10, 2026

66.9Accuracy

7

Feb 27, 2026

SEED-Bench (test)

52.38TU (Text Understanding)

5

May 19, 2026

85.5Final Performance

5

Apr 7, 2026

62.9Final Performance

5

Apr 7, 2026

2,392MME-S Score

5

Mar 4, 2026

60.5Accuracy

4

Apr 21, 2026

83.9Performance Score

3

May 21, 2026

52.5S-I-U

3

Feb 26, 2026

LLaVA Multi-modal Evaluation Suite (GQA, MMB, MME, POPE, SQA, VQAv2, TextVQA, MMMU, SEED-I) v1.6 (test val)

—

—GQA

0

Feb 26, 2026

Page 2 of 2

Popular datasets

MMBench MME MMMU SEED-Bench SEED

Follow for updates

Join SOTA2 Contact