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

Video Understanding on MotionBench (val)

65.4Accuracy

GPT-5

Updated 18d ago

Evaluation Results

Method	Links
GPT-5 2026.01		65.4
GPT-5 2026.03		65.4
Gemini 3 Pro 2026.01		62.6
Gemini 3 Pro 2026.03		62.6
Molmo2-8B 2026.01		62.2
Molmo2-8B 2026.03		62.2
Gemini 2.5 Pro 2026.01		62
Gemini 2.5 Pro 2026.03		62
Molmo2-4B 2026.01		61.6
Molmo2-4B 2026.03		61.6
PLM-8B 2026.01		61.4
PLM-8B 2026.03		61.4
PLM-8B 2026.03		61.4
MolmoPoint-8B 2026.03		61.4
Baseline 2026.03		61
VideoChat-Flash-7B 2026.01		60.6
Molmo2-O-7B 2026.01		60.6
VideoChat-Flash-7B 2026.03		60.6
Molmo2-O-7B 2026.03		60.6
PLM-3B 2026.01		60.4
PLM-3B 2026.03		60.4
GPT-5 mini 2026.01		59.9
GPT-5 mini 2026.03		59.9
MiniCPM-V-4.5-8B 2026.01		59.7
MiniCPM-V-4.5-8B 2026.03		59.7
Gemini 2.5 Flash 2026.01		59.3
Gemini 2.5 Flash 2026.03		59.3
STTS 2026.03		59.2
STTS 2026.03		59.2
GLM-4.1V-9B 2026.01		59
GLM-4.1V-9B 2026.03		59
Qwen3-VL-4B 2026.01		58.6
Qwen3-VL-4B 2026.03		58.6
Qwen3-VL-4B 2026.03		58.6
Claude Sonnet 4.5 2026.01		58.5
Claude Sonnet 4.5 2026.03		58.5
STTS 2026.03		58.2
Qwen3-VL-8B 2026.01		56.9
Qwen3-VL-8B 2026.03		56.9
InternVL3.5-8B 2026.01		56.6
InternVL3.5-8B 2026.03		56.6
InternVL3.5-8B 2026.03		56.6
InternVL3.5-4B 2026.01		56.5
InternVL3.5-4B 2026.03		56.5
Eagle2.5-8B 2026.01		55.7
Eagle2.5-8B 2026.03		55.7
Keye-VL-1.5-8B 2026.01		55.1
Keye-VL-1.5-8B 2026.03		55.1
LLaVA-Video-7B 2026.01		54.2
LLaVA-Video-7B 2026.03		54.2