List your company

Visual Search and Perception-intensive Reasoning on HR-Bench 4K

76.63Overall ScoreStarve to Perceive
36.454846.884957.31567.7451May 18, 2026

Evaluation results

MethodLinks
Starve to PerceiveVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.0576.63
Qwen2.5-VL + BA-SFT + VanillaRLVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.0575.25
Qwen2.5-VL + BA-SFTVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.0572.5
ChainOfFocusVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.0571.63
DeepEyesVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.0571.5
PixelReasonerVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.0571.13
GPT-4oVisual Bandwidth Constraint (B)=No Constrain, Reported From=Other publications2026.0568
Qwen2.5-VL + BA-SFT + VanillaRLVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.0566.25
Starve to PerceiveVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.0566.13
Qwen2.5-VLVisual Bandwidth Constraint (B)=No Constrain, Reported From=Other publications2026.0565.25
LLaVA-OneVisionVisual Bandwidth Constraint (B)=No Constrain, Reported From=Other publications2026.0563
ChainOfFocusVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.0559.63
Qwen2.5-VL + BA-SFTVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.0559.25
Mini-o3Visual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.0557.75
DeepEyesVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.0557.13
Mini-o3Visual Bandwidth Constraint (B)=256, Tool-calling=True2026.0551.63
PixelReasonerVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.0549.38
Qwen2.5-VLVisual Bandwidth Constraint (B)=2562026.0538