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

Multimodal Reasoning on ERQA

55Accuracy

Best Model

Updated 12d ago

Evaluation Results

Method	Links
Best Model 2026.05		55
MAESTRO* 2026.05		52.5
Qwen3-VL-32B 2026.05		49.3
Direct Answering 2026.05		47.5
Thyme 2026.05		47.3
Kimi-K2.5 2026.05		46
GPT-5 2026.05		45.8
GPT-4o 2026.05		44.8
Gemini-2.5-Flash 2026.05		44.3
GLM-4.6V 2026.05		44
Gemini-2.5-Pro 2026.05		43
MAESTRO 2026.05		42.8
DeepEyes-v2 2026.05		42.3
Chain-of-Focus 2026.05		41.8
Untrained Model 2026.05		41.5
VisionReasoner 2026.05		39.3
VTS-V 2026.05		37.5
MathCoder-VL 2026.05		37.3
Visual-ARFT 2026.05		37
VTOOL-R1 2026.05		36.8
DeepEyes 2026.05		35.5
PixelReasoner 2026.05		33.5