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

Image editing point-wise evaluation on ImagenHub

41.84Spearman Rank Correlation

Human-to-Human

Updated 23d ago

Evaluation Results

Method	Links
Human-to-Human 2025.09		41.84
Gemini-2.5-Flash 2025.09		41.62
GPT-5 2025.09		40.85
GPT-4o 2025.09		38.21
EDITREWARD 2025.09		36.18
EDITREWARD 2025.09		35.2
ADIEE 2025.09		34.5
Qwen2.5-VL-32B-Inst 2025.09		26.87
Gemini-2.0-Flash 2025.09		23.69
MiMo-VL-7B-SFT-2508 2025.09		22.14
Qwen2.5-VL-7B-Inst 2025.09		18.59
ReasonEdit 2026.05		0.7566
Gemini-2.5-Flash 2026.05		0.4162
GPT-5 2026.05		0.4085
GPT-4o 2026.05		0.3821
LMM4Edit 2026.05		0.3726
EditReward (MiMo) 2026.05		0.352
EditScore (Qwen3) 2026.05		0.3062
Qwen3-VL (7B) 2026.05		0.2426
Gemini-2.0-Flash 2026.05		0.2369
InternVL3.5 (7B) 2026.05		0.2071
Qwen2.5-VL-3B-Inst 2025.09		-2.54