SOTA Long-form Video Question Answering on EgoSchema and PapersWithCode

77.9Accuracy

Qwen2.5-VL-72B

Updated 1mo ago

Evaluation Results

Method	Links
Qwen2.5-VL-72B 2025.06		77.9
Human Performance 2024.04		76.2
GPT-4o(0513) 2025.06		72.2
VideoChat-A1 2025.06		72.1
Gemini 1.5 Pro 2025.06		71.1
VideoChat-A1 2025.06		70.7
VideoChat-A1 2025.06		70.1
VideoMultiAgents 2025.06		68
LongVU-7B 2025.06		67.6
LLAVA-Video-72B 2025.06		65.6
Qwen2.5-VL-7B 2025.06		65
VideoLLaMA-2-72B 2025.06		63.9
InternVideo-2.5-8B 2025.06		63.9
LLaVA-OneVision-72B 2025.06		62
VideoTree 2025.06		61.1
DrVideo 2025.06		61
LLaVA-Video-7B 2025.06		57.3
VideoChat2-7B 2025.06		56.7
MovieChat+ 2024.04		56.4
VideoAgent 2025.06		54.1
MovieChat 2024.04		53.5
Video Chat 2024.04		47.5
FrozenBiLM 2024.04		26.9
Choosing the correct answer uniformly at random 2024.04		20