List your company

Long Video Understanding on Video-MME long 1.0

67.4Accuracy (No Subs)Gemini 1.5 Pro
33.70442.45251.259.948Jun 18, 2024Jul 16, 2024Aug 14, 2024Sep 11, 2024Oct 10, 2024Nov 7, 2024Dec 6, 2024

Evaluation results

MethodLinks
Gemini 1.5 ProFrames=0.5 FPS2024.0667.477.4
Gemini 1.5 Pro2024.1267.477.4
Gemini-1.5-Pro2024.0967.477.4
GPT-4oFrames=3842024.0665.372.1
GPT4-O2024.1265.372.1
GPT-4o2024.0965.372.1
LinVT-Qwen2-VLSize=7B2024.1263.163.3
Qwen2-VL-72BFrames=7682024.0662.274.3
Gemini 1.5 FlashFrames=0.5 FPS2024.0661.168.8
LinVT-InternVL2Size=8B2024.1260.263.5
LLaVA-OneVision-72BFrames=322024.066062.4
LLaVA-OneVisionSize=72B2024.096062.4
Oryx-1.5Size=32B2024.0959.169.7
GPT-4o miniFrames=2502024.0658.663.4
LinVT-BLIP-3Size=4B2024.1258.362.4
LinVT-MolmoSize=7B2024.1257.661.2
OryxSize=34B2024.0953.958
VILA-1.5Size=40B2024.0953.855.7
GPT-4VFrames=102024.0653.556.9
GPT4-V2024.1253.556.9
GPT-4V2024.0953.556.9
LinVT-AquilaSize=2B2024.1253.455.3
DrVideoFrames=0.2 FPS2024.0651.771.7
Claude 3.5 SonnetFrames=202024.0651.254.7
Oryx-1.5Size=7B2024.0951.258.3
OryxSize=7B2024.1250.355.8
OryxSize=7B2024.0950.355.8
VITA-8x7BFrames=322024.0648.650.9
VideoCCAMSize=14B2024.1246.749.9
VideoCCAMSize=14B2024.0946.749.9
KangarooSize=8B2024.1246.649.3
KangarooSize=8B2024.0946.649.3
LongVAFrames=1282024.0646.247.6
LongVASize=7B2024.1246.247.6
LongVASize=7B2024.0946.247.6
LLoViFrames=0.2 FPS, re-implemented=true2024.0645.467.7
LinVT-MiphaSize=1.6B2024.1244.546.1
VideoLLaMA2Size=7B2024.1242.143.8
VideoLLaMA2Size=7B2024.0942.143.8
VideoAgentFrames=7.6, re-implemented=true2024.0640.244.4
Video-LLaVASize=7B2024.1239.941.6
LongVILASize=8B2024.1239.7—
LongVILASize=8B2024.0939.7—
Qwen-VL-ChatSize=7B2024.1237.837.9
ShareGPT4VideoSize=8B2024.123537.9
DrVideo (only subs)Frames=0.2 FPS, subtitles_only=true2024.06—68.5