Visual Spatial Reasoning on VSR
88.87AccuracyQwen3-VL-32B
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3-VL-32BToken Ratio=100%2026.05 | 88.87 | |
| CDPrunerToken Ratio=60%2026.05 | 88.46 | |
| VisionZipToken Ratio=60%2026.05 | 88.38 | |
| F3AToken Ratio=60%2026.05 | 88.22 | |
| DivPruneToken Ratio=60%2026.05 | 87.89 | |
| FastVToken Ratio=60%2026.05 | 87.15 | |
| VisionZipToken Ratio=40%2026.05 | 86.58 | |
| F3AToken Ratio=40%2026.05 | 86.58 | |
| CDPrunerToken Ratio=40%2026.05 | 86.5 | |
| DivPruneToken Ratio=40%2026.05 | 86.25 | |
| InternVL3.5-38BRatio=100%, Backbone=InternVL3.5-38B2026.05 | 84.04 | |
| F3AToken Ratio=20%2026.05 | 84.04 | |
| FastVToken Ratio=40%2026.05 | 83.8 | |
| CDPrunerToken Ratio=20%2026.05 | 83.72 | |
| F3ARatio=60%, Backbone=InternVL3.5-38B2026.05 | 83.7 | |
| UniWorld-v1Params=7B+12B‡2026.05 | 83.34 | |
| FastVRatio=60%, Backbone=InternVL3.5-38B2026.05 | 83.32 | |
| F3ARatio=40%, Backbone=InternVL3.5-38B2026.05 | 83.28 | |
| VisionZipRatio=60%, Backbone=InternVL3.5-38B2026.05 | 83.22 | |
| VisionZipToken Ratio=20%2026.05 | 83.22 | |
| CDPrunerRatio=60%, Backbone=InternVL3.5-38B2026.05 | 83.04 | |
| DivPruneRatio=60%, Backbone=InternVL3.5-38B2026.05 | 82.72 | |
| FastVRatio=40%, Backbone=InternVL3.5-38B2026.05 | 82.53 | |
| DivPruneToken Ratio=20%2026.05 | 82.41 | |
| CDPrunerRatio=40%, Backbone=InternVL3.5-38B2026.05 | 82.36 | |
| DivPruneRatio=40%, Backbone=InternVL3.5-38B2026.05 | 82.36 | |
| VisionZipRatio=40%, Backbone=InternVL3.5-38B2026.05 | 81.83 | |
| Qwen2.5-VL-7BRatio=100%, Backbone=Qwen2.5-VL-7B2026.05 | 81.66 | |
| SGT-BAGELParams=7B+7B‡2026.05 | 81.54 | |
| F3ARatio=60%, Backbone=Qwen2.5-VL-7B2026.05 | 81.25 | |
| CDPrunerRatio=60%, Backbone=Qwen2.5-VL-7B2026.05 | 81.01 | |
| F3ARatio=20%, Backbone=InternVL3.5-38B2026.05 | 80.87 | |
| FastVRatio=60%, Backbone=Qwen2.5-VL-7B2026.05 | 80.84 | |
| DivPruneRatio=60%, Backbone=Qwen2.5-VL-7B2026.05 | 80.84 | |
| F3ARatio=40%, Backbone=Qwen2.5-VL-7B2026.05 | 80.7 | |
| BAGELParams=7B+7B‡2026.05 | 80.45 | |
| DivPruneRatio=40%, Backbone=Qwen2.5-VL-7B2026.05 | 80.19 | |
| FastVRatio=20%, Backbone=InternVL3.5-38B2026.05 | 80.12 | |
| CDPrunerRatio=40%, Backbone=Qwen2.5-VL-7B2026.05 | 80.03 | |
| FastVRatio=40%, Backbone=Qwen2.5-VL-7B2026.05 | 80.03 | |
| DivPruneRatio=20%, Backbone=InternVL3.5-38B2026.05 | 79.84 | |
| CDPrunerRatio=20%, Backbone=InternVL3.5-38B2026.05 | 79.42 | |
| VisionZipRatio=20%, Backbone=InternVL3.5-38B2026.05 | 79.3 | |
| VisionZipRatio=60%, Backbone=Qwen2.5-VL-7B2026.05 | 79.27 | |
| VisionZipRatio=40%, Backbone=Qwen2.5-VL-7B2026.05 | 79.22 | |
| ActiveScopeModel=Qwen3-VL 4B2026.06 | 79 | |
| SGT-Gen2Params=3B+4B‡2026.05 | 78.85 | |
| ZoomEyeModel=Qwen3-VL 4B2026.06 | 78.66 | |
| F3ARatio=20%, Backbone=Qwen2.5-VL-7B2026.05 | 78.48 | |
| RegularModel=Qwen3-VL 4B2026.06 | 78.14 | |
| VisionZipRatio=20%, Backbone=Qwen2.5-VL-7B2026.05 | 77.99 | |
| CDPrunerRatio=20%, Backbone=Qwen2.5-VL-7B2026.05 | 77.58 | |
| OmniGen2Params=3B+4B‡2026.05 | 77.52 | |
| DivPruneRatio=20%, Backbone=Qwen2.5-VL-7B2026.05 | 77.41 | |
| FastVToken Ratio=20%2026.05 | 77.41 | |
| FastVRatio=20%, Backbone=Qwen2.5-VL-7B2026.05 | 75.13 | |
| UniMRGParams=3.6B2026.05 | 73.9 | |
| Janus-ProParams=7B2026.05 | 71.03 | |
| OpenUniParams=2B2026.05 | 66.69 | |
| UniLIPParams=2B2026.05 | 65.55 | |
| HarmonParams=1.5B2026.05 | 60.88 | |
| Show-o512Params=1.3B2026.05 | 54.26 |