List your company

Video Understanding on MMBench-Video

1.87Mean Score (0-3)GPT-4o (2024-05-13)
1.57881.65441.731.8056Nov 14, 2025Dec 7, 2025Dec 31, 2025Jan 24, 2026Feb 17, 2026Mar 13, 2026Apr 6, 2026

Evaluation results

MethodLinks
GPT-4o (2024-05-13)Size=-2026.041.87—
G2F-RAGSize=7B, Base Model=Qwen2.5-VL2026.041.85—
G2F-RAGSize=8B, Base Model=InternVL3.52026.041.82—
SlowFast + PASInference Strategy=SlowFast stacked with PAS2025.111.81—
Gemini1.5 ProSize=-2026.041.8—
G2F-RAGSize=14B, Base Model=InternVL3.52026.041.8—
Qwen2.5-VLSize=7B2026.041.79—
PASInference Strategy=Phase Aggregated Smoothing (Ours)2025.111.78—
SlowFast-LLaVAInference Strategy=SlowFast2025.111.76—
TS-LLaVA + PASInference Strategy=TS-LLaVA stacked with PAS2025.111.76—
InternVL3.5Size=8B2026.041.74—
InternVL3Size=14B2026.041.73—
InternVL3.5Size=14B2026.041.73—
Default SettingInference Strategy=Default Setting, Backbone=Qwen2.5-VL-7B-Instruct2025.111.71—
TS-LLaVAInference Strategy=TS-LLaVA2025.111.7—
MiniCPM-V 2.6Size=8B2026.041.7—
InternVL3Size=8B2026.041.69—
InternVL2.5Size=8B2026.041.68—
G2F-RAGSize=4B, Base Model=InternVL3.52026.041.66—
G2F-RAGSize=7B, Base Model=LLaVA-Video2026.041.66—
Qwen2.5-VLSize=3B2026.041.63—
LLaVA-VideoSize=7B2026.041.6—
InternVL3.5Size=4B2026.041.59—
GPT-4o2025.06—2.15
KangarooSize=8B2024.12—1.44
LinVT-AquilaSize=2B2024.12—1.12
LinVT-BLIP-3Size=4B2024.12—1.53
LinVT-InternVL2Size=8B2024.12—1.57
LinVT-MiphaSize=1.6B2024.12—1.06
LinVT-MolmoSize=7B2024.12—1.54
LinVT-Qwen2-VLSize=7B2024.12—1.62
MiniGPT4-VideoSize=7B2024.12—1.05
PLLAVASize=7B2024.12—1.03
UnisonStage=Uni. 2 Stage2025.12—0.94
UniWorld-V12025.06—1.79
Video-LLaVA2025.06—1.05
Video-LLaVASize=7B2024.12—1.05
VideoChatGPTSize=7B2024.12—0.93
VideoLLaMA 2# Frames=16, Parameters=7B2024.06—1.06