List your company

Audiovisual Dialogue Description on DiaDemBench

97.9REFHuman (avg. of 3 authors)
8.04431.37254.778.028Jan 27, 2026

Evaluation results

MethodLinks
Human (avg. of 3 authors)2026.0197.997.1
DiaDemSize=7B2026.0165.979.3
Gemini-2.5-Pro2026.0163.674.8
Gemini-3-Pro2026.0163.171
Gemini-2.5-Flash2026.0158.573.3
Qwen3-Omni-CaptionerSize=30B-A3B2026.0143.958.8
AVoCaDOSize=7B2026.0138.751.7
Qwen3-Omni-InstructSize=30B-A3B2026.0136.847.5
ARC-Qwen-Video-NarratorSize=7B2026.0132.848.5
UGC-VideoCaptionerSize=3B2026.0129.747
Qwen2.5-OmniSize=7B2026.0126.137.1
OmniVinciSize=9B2026.0117.125.2
HumanOmniV2Size=7B2026.0115.321.5
ARC-Qwen-VideoSize=7B2026.0111.717.2
video-SALMONN-2Size=7B2026.0111.516.6