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

Visual Search on V* Bench (Accuracy)

90.4Accuracy

Deepeyes-7B

Updated 16d ago

Evaluation Results

Method	Links
Deepeyes-7B 2025.12		90.4
Multi-t ViGoRL-7B 2025.05		86.4
Qwen2.5-VL-32B 2025.12		85.9
Qwen2.5-VL-72B 2025.12		84.8
CodeDance-7B 2025.12		84.8
Pixel Reasoner-7B 2025.12		84.3
Thyme-VL-7B 2025.12		82.2
IVM-Enhanced GPT-4V 2025.05		81.2
Multi-t ViGoRL-3B 2025.05		81.2
Multi-t ViGoRL-3B 2025.05		81.2
Sketchpad-GPT-4o 2025.05		80.3
ViGoRL-3B (Ours) 2025.05		79.1
Qwen2.5-7B-VL 2025.05		78
Multi-t ViGoRL-3B 2025.05		78
InternVL3-78B 2025.12		76.4
Qwen2.5-VL-7B 2025.12		76.4
SEAL 2025.05		74.8
Qwen2.5-3B-VL 2025.05		74.2
Llava-OneVision-72B 2025.12		73.8
InternVL2.5-8B 2025.12		73.7
Qwen2.5-VL-7B +FINER-Tuning 2026.03		72.8
Llava-OneVision-7B 2025.12		72.7
Qwen2.5-VL-7B 2026.03		72.7
InternVL3.5-8B +FINER-Tuning 2026.03		71.2
InternVL3-8B 2025.12		70.2
InternVL3.5-14B +FINER-Tuning 2026.03		70.2
InternVL3.5-8B 2026.03		69.1
InternVL3.5-14B 2026.03		68
GPT-4o 2025.12		67.5
GPT-4o 2025.05		66
LLaVA-1.6-13B 2025.05		61.8
LLaVA-1.6-7B +FINER-Tuning 2026.03		55
GPT-4V 2025.05		55
OmniLMM-12B +RLAIF-V 2026.03		54.4
LLaVA-1.6-7B 2026.03		53.9
OmniLMM-12B 2026.03		52.9
LLaVA-1.5-7B 2025.05		48.7
Gemini-Pro 2025.05		48.2
VisProg 2025.05		41.4
MM-React 2025.05		41.4
VisualChatGPT 2025.05		37.6