Multi-view spatial reasoning on MINDCUBE full
58.56Overall AccuracypySpatial
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| pySpatialcode agent=GPT-4.1-mini2026.03 | 58.56 | 43.2 | 60.54 | 48.1 | |
| DeepSeek-VL2-Small2026.03 | 47.62 | 37 | 50.38 | 26.91 | |
| LLaVA-OneVision-7B2026.03 | 47.43 | 36.45 | 48.42 | 44.09 | |
| GPT-4.1-mini2026.03 | 45.62 | 37.84 | 47.22 | 34.56 | |
| mPLUG-Owl3-7B-2411012026.03 | 44.85 | 37.84 | 47.11 | 26.91 | |
| Claude-4-Sonnet2026.03 | 44.75 | 48.42 | 44.21 | 47.62 | |
| VLM-3R2026.03 | 42.09 | 36.73 | 44.22 | 24.45 | |
| LLaVA-Video-Qwen-7B2026.03 | 41.96 | 35.71 | 43.55 | 30.12 | |
| GPT-4o2026.03 | 38.81 | 32.65 | 40.17 | 29.16 | |
| RoboBrain2026.03 | 37.38 | 35.8 | 38.28 | 29.53 | |
| Space-Qwen2026.03 | 33.28 | 38.02 | 33.71 | 26.32 | |
| Qwen2.5-VL-3B-Instruct2026.03 | 33.21 | 37.37 | 33.26 | 30.34 | |
| Random (frequency)2026.03 | 33.02 | 38.3 | 32.66 | 35.79 | |
| Random (chance)2026.03 | 32.35 | 36.36 | 32.29 | 30.66 | |
| Spatial-MLLM2026.03 | 32.06 | 38.39 | 20.92 | 32.82 | |
| Qwen2.5-VL-7B-Instruct2026.03 | 29.26 | 38.76 | 29.5 | 21.35 | |
| SpaceMantis2026.03 | 22.81 | 37.65 | 21.26 | 29.32 | |
| InternVL2.5-8B2026.03 | 18.68 | 36.45 | 18.2 | 13.11 |