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

Visual Search and Perception-intensive Reasoning on Multi-benchmark Score Overall

64.59Overall Score

Starve to Perceive

Updated 15d ago

Evaluation Results

Method	Links
Starve to Perceive 2026.05		64.59
Qwen2.5-VL + BA-SFT + VanillaRL 2026.05		64.48
DeepEyes 2026.05		61.77
ChainOfFocus 2026.05		61.75
PixelReasoner 2026.05		60.1
Qwen2.5-VL + BA-SFT 2026.05		58.95
Qwen2.5-VL 2026.05		52.53
Starve to Perceive 2026.05		52.35
Qwen2.5-VL + BA-SFT + VanillaRL 2026.05		51.54
Mini-o3 2026.05		46.68
Qwen2.5-VL + BA-SFT 2026.05		46.6
DeepEyes 2026.05		45.57
ChainOfFocus 2026.05		45.25
Mini-o3 2026.05		36.75