Spatial Reasoning on SPAR-Bench SI, MV 91
63.3Accuracy3DThinker-S1+S2
Evaluation results
| Method | Links | |
|---|---|---|
| 3DThinker-S1+S2Base Model=Qwen2.5-VL-7B, Stage=1+22025.10 | 63.3 | |
| 3DThinker-S1+S2Base Model=Qwen2.5-VL-3B, Stage=1+22025.10 | 58.2 | |
| 3DThinker-S1Base Model=Qwen2.5-VL-7B, Stage=12025.10 | 56.3 | |
| 3DThinker-S1Base Model=Qwen2.5-VL-3B, Stage=12025.10 | 52.3 | |
| VILASR-7BBase Model=Qwen2.5-VL-7B2025.10 | 37.8 | |
| SpaceR-7BBase Model=Qwen2.5-VL-7B2025.10 | 37.1 | |
| Spatial-MLLM-4BBase Model=Qwen2.5-VL-3B2025.10 | 35.1 | |
| SpatialLadder-3BBase Model=Qwen2.5-VL-3B2025.10 | 34.4 | |
| Video-R1Base Model=Qwen2.5-VL-7B2025.10 | 31.5 | |
| Qwen2.5-VL-7BBase Model=Qwen2.5-VL-7B2025.10 | 30.2 | |
| Qwen2.5-VL-3BBase Model=Qwen2.5-VL-3B2025.10 | 24.6 |