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

Text-to-Video Generation research benchmarks – Page 9 · SOTA2 Research

Text-to-Video Generation

Benchmarks

Dataset NameSOTA MethodSort

Dataset NameSOTA methodMetricTrendResultsLast Updated

23.58FID

2

May 12, 2026

121-frame video generation prompts (test)

26.32PSNR

2

Apr 28, 2026

VBench Wan2.2-TI2V-5B

0.584Total Score

2

Mar 4, 2026

Semantic Erasure and Horizontal Injection

0.664Dynamic Scenes Score

2

Feb 26, 2026

Movie 81 frames, 256x256 (val)

40.8UR

2

Feb 26, 2026

MovieGenBench 81 frames, 256x256

29.8UR

2

Feb 26, 2026

VBench 81 frames, 256x256

35.9UR

2

Feb 26, 2026

Human Study 4K Video Generation (test)

71.25Video Quality

2

Feb 26, 2026

80.33Score

2

Feb 26, 2026

VMBench motion-oriented prompts 14

65.02Aesthetic Quality

2

Feb 26, 2026

Our Eval. Set (300 prompts)

77.15Quality

2

Feb 26, 2026

DrawBench 200 prompts

0.769Quality

2

Feb 26, 2026

Multimodal VoxCeleb

0.197CLIP Score

2

Feb 26, 2026

209.127FVD

2

Feb 26, 2026

ACDC & DSB (test)

Oracle baseline

19.41FVD (Real vs Real)

1

Jun 16, 2026

OmniCAM Reference-video-controlled

868.63FVD

1

Apr 8, 2026

Concerts dataset

sequential two-step text-to-audio-video generation pipeline

5,020.37FAD

1

Mar 18, 2026

VDM prompts (28 examples)

84.38Quality

1

Feb 26, 2026

50 video prompts (test)

—

—Primary metric

0

Feb 26, 2026

Page 9 of 9

Popular datasets

VBench Simulated Videos Wan HunyuanVideo User Study

Follow for updates

Join SOTA2 Contact