Perception-intensive Reasoning on MME-RealWorld-Lite
55.13ScoreStarve to Perceive
Evaluation Results
| Method | Links | |
|---|---|---|
| Starve to PerceiveVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.05 | 55.13 | |
| Qwen2.5-VL + BA-SFT + VanillaRLVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.05 | 53.78 | |
| DeepEyesVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.05 | 50.81 | |
| ChainOfFocusVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.05 | 48.98 | |
| Starve to PerceiveVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.05 | 47.37 | |
| PixelReasonerVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.05 | 47.21 | |
| Qwen2.5-VL + BA-SFTVisual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.05 | 46.95 | |
| Qwen2.5-VL + BA-SFT + VanillaRLVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.05 | 46.69 | |
| GPT-4oVisual Bandwidth Constraint (B)=No Constrain, Reported From=Other publications2026.05 | 46.4 | |
| LLaVA-OneVisionVisual Bandwidth Constraint (B)=No Constrain, Reported From=Other publications2026.05 | 43.7 | |
| Qwen2.5-VLVisual Bandwidth Constraint (B)=No Constrain, Reported From=Other publications2026.05 | 43 | |
| DeepEyesVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.05 | 40.85 | |
| Qwen2.5-VL + BA-SFTVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.05 | 39.55 | |
| ChainOfFocusVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.05 | 37.83 | |
| Mini-o3Visual Bandwidth Constraint (B)=No Constrain, Tool-calling=True2026.05 | 32.15 | |
| PixelReasonerVisual Bandwidth Constraint (B)=256, Tool-calling=True2026.05 | 29.6 | |
| Qwen2.5-VLVisual Bandwidth Constraint (B)=2562026.05 | 27.41 | |
| Mini-o3Visual Bandwidth Constraint (B)=256, Tool-calling=True2026.05 | 27.36 |