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

VLM-as-a-Judge on MLLM-as-a-Judge

75.78Accuracy

GPT-4o

Updated 1mo ago

Evaluation Results

Method	Links
GPT-4o 2026.04		75.78
Llama-3.2-Vision-90B 2026.04		75.12
Gemini-2.5-Flash 2026.04		73.37
Gemini-2.5-Pro 2026.04		73.37
LLaVA-OneVision-1.5-8B 2026.04		71.17
Gemini-2.5-Pro 2026.04		69.21
Gemini-2.5-Flash 2026.04		68.34
Gemini-2.5-Pro 2026.04		68.04
Gemini-2.5-Pro 2026.04		67.63
Gemini-2.5-Flash 2026.04		66.92
Gemini-2.5-Flash 2026.04		66.9
Llama-3.2-Vision-90B 2026.04		66.73
GPT-4o 2026.04		66.45
LLaVA-OneVision-1.5-8B 2026.04		66.21
GPT-4o 2026.04		65.79
LLaVA-OneVision-1.5-8B 2026.04		65.67
LLaVA-OneVision-1.5-8B 2026.04		65.35
Llama-3.2-Vision-90B 2026.04		65.28
GPT-4o 2026.04		65.14
Llama-3.2-Vision-90B 2026.04		63.18
Phi-4-multimodal 2026.04		63.11
LLaVA-1.6-34B 2026.04		61.54
LLaVA-1.5-13B 2026.04		59.48
Phi-4-multimodal 2026.04		59.19
Phi-4-multimodal 2026.04		59.11
Phi-4-multimodal 2026.04		59.07
LLaVA-1.6-34B 2026.04		58.95
LLaVA-1.6-34B 2026.04		58.88
LLaVA-1.6-34B 2026.04		58.02
LLaVA-1.5-13B 2026.04		56.21
LLaVA-1.5-13B 2026.04		54.69
LLaVA-1.5-13B 2026.04		53.28