List your company

Video Understanding on LSDBench (test)

56.2AccuracyGemini-2.0-Flash
30.30437.02743.7550.473Mar 24, 2026

Evaluation results

MethodLinks
Gemini-2.0-FlashFrames=2700, Visual Token=696.6k2026.0356.2
Qwen2.5-VLFrames=768, Visual Token=499.2k2026.0352.5
Qwen2.5-VL(RHS)Frames=225, Visual Token=146.2k2026.0352.2
EVAFrames=76.9, Visual Token=10.3k2026.0351
Qwen2.5-VLFrames=256, Visual Token=166.4k2026.0350.1
LongVilaFrames=256, Visual Token=65.5k2026.0349.8
Qwen2.5-VL*Frames=32, Visual Token=21.0k2026.0349.2
Qwen2-VLFrames=256, Visual Token=166.4k2026.0348
Qwen2.5-VL*Frames=16, Visual Token=10.5k2026.0347.7
LongVAFrames=512, Visual Token=73.7k2026.0333
LongVAFrames=1024, Visual Token=147.5k2026.0332.5
LongVAFrames=256, Visual Token=36.9k2026.0331.3