List your company
ResearchFormerly WizWand

Long-context video understanding on LongVideoBench (test)

66.7AccuracyGPT-4o
42.57248.83655.161.364Oct 8, 2024Oct 17, 2024Oct 27, 2024Nov 6, 2024Nov 15, 2024Nov 25, 2024Dec 5, 2024
Updated 8mo ago

Evaluation results

MethodLinks
GPT-4ocategory=Proprietary2024.1066.7
GPT-4o2024.1066.7
GPT-4o2024.1266.7
ARIAactivated_params=3.9B (25.3B), category=Open-source2024.1065.3
ARIA2024.1065.3
Gemini-1.5-Procategory=Proprietary2024.1064.4
Gemini-1.5 Pro2024.1064.4
LLaVA-OneVision-72Bactivated_params=72B, category=Open-source2024.1063.2
Gemini-1.5-Flashcategory=Proprietary2024.1062.6
Gemini-1.5 Flash2024.1062.4
Qwen2-VL-72Bactivated_params=72B, category=Open-source2024.1061.7
GPT-4V2024.1060.7
InternVL2-40Bactivated_params=40B, category=Open-source2024.1060.6
GPT-4o minicategory=Proprietary2024.1058.8
GPT-4o mini2024.1058.8
GPT-4o mini2024.1258.8
NVILANumber of Parameters=8B, #Frames=2562024.1258.7
Qwen2-VL-7Bactivated_params=7B, category=Open-source2024.1056.8
Qwen2-VLNumber of Parameters=8B, #Frames=2fps2024.1256.8
MiniCPM-V-2.6activated_params=8B, category=Open-source2024.1055.7
InternVL-Chat-V1.5activated_params=26B, category=Open-source2024.1051.7
Video-XLNumber of Parameters=7B, #Frames=20482024.1251.3
Idefics2activated_params=8B, category=Open-source2024.1049.4
Pixtral-12Bactivated_params=12B, category=Open-source2024.1047.4
Pixtral-12B2024.1047.4
Llama3.2-11Bactivated_params=11B, category=Open-source2024.1045.7
Llama3.2-11B2024.1045.7
LLaVA-NeXT-VideoNumber of Parameters=7B, #Frames=322024.1243.5