Share your thoughts, 1 month free Claude Pro on usSee more

Long Video Understanding on LongVideoBench (Accuracy)

67Accuracy

Penguin-VL

Updated 3d ago

Evaluation Results

Method	Links
Penguin-VL 2026.03		67
GPT-4o 2026.04		66.7
GPT-4o 2026.04		66.7
Eagle2.5-8B 2026.04		66.4
VideoChat-Flash-7B 2026.04		64.7
Gemini-1.5-Pro 2026.04		64
RLER 2026.04		63
Qwen3-VL 2026.03		62.6
InternVL3-9B 2026.04		62.5
InternVL-3.5 2026.03		62.1
mPLUG-Owl3-8B 2026.04		59.8
XComp 2026.04		59.7
LLaVA-Video-7B 2026.04		59.5
GPT-4V 2026.04		59.1
VideoChat-Flash-2B 2026.04		58.3
TimeMarker 2026.04		56.3
Qwen2.5-VL-7B 2026.04		56
InternVL3-2B 2026.04		55.4
Kangaroo 2026.04		54.8
mPLUG-Owl3 2026.04		52.1
Qwen2.5VL-3B 2026.04		43.3
ShareGPT4Video-8B 2026.04		39.7
OpenAI GPT-5 nano 2026.03		38.1