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

General Visual Reasoning on MMStar (Accuracy)

77.5Accuracy

Gemini 2.5 Pro

Updated 19d ago

Evaluation Results

Method	Links
Gemini 2.5 Pro 2026.02		77.5	-
InternVL3-8B 2025.09		66.3	-
SAYO-Qwen-8B 2026.02		65.27	-
GPT4o 2026.02		64.7	-
GPT-4o 2026.01		64.7	-
GPT-4o 2025.09		64.7	53.2
MM-Eureka 2025.09		64	52.52
NoisyRollout-7B 2026.02		63.67	-
SAYO-Qwen-4B 2026.02		63.53	-
Kimi-VL-16B 2026.02		63.47	-
VL-Rethinker 2026.01		63.2	-
OpenVLThinker-7B 2026.02		63.07	-
Semantic-back-7B 2026.02		63.07	-
AdaVaR-7B 2025.09		63	55.82
SAYO-InternVL-8B 2026.02		62.73	-
OVR-7B 2025.09		62.7	55.76
Vision-R1 2026.01		62.67	-
Qwen3-VL-8B 2026.02		62.6	-
VLAA-Thinker-7B 2025.09		62.6	54.2
ViGoRL 2026.02		61.53	-
DeepEyes 2025.09		61.3	53.72
Qwen3-VL-4B 2026.02		60.8	-
Monet 2026.01		60.33	-
Qwen2.5-VL-7B 2025.09		60.3	50.9
InternVL3.5-8B 2026.02		60.27	-
Laser 2026.01		60.27	-
Qwen2.5-VL-7B 2026.01		59.7	-
Orsta-7B 2025.09		59.6	52.04
AdaVaR-3B 2025.09		59.3	50.84
Chain-of-Focus 2025.09		59.3	50.94
LLaVA-OneVision 2026.01		59.13	-
DeepEyes 2026.01		58.73	-
InternVL3.5-30B-A3B 2026.02		58.33	-
LVR 2026.01		57.93	-
R1-Onevision-7B 2026.02		57	-
InternVL3.5-14B 2026.02		55.93	-
LMM-R1 2025.09		55	48.8
VLAA-Thinker-3B 2025.09		54.8	46.7
ViGoRL-7B 2025.09		54.3	50.53
InternVL3.5-38B 2026.02		54.2	-
Qwen3-VL-30B-A3B 2026.02		53.73	-
InternVL3.5-8B 2026.01		53.33	-
Qwen2.5-VL-3B 2025.09		53.1	45.74
GRIT-3B 2025.09		52.7	45.8
ViGoRL-3B 2025.09		51.4	45.7
PAPO 2026.01		45.8	-