Spatial Reasoning on 6-DoF SpatialBench 41
66Relative Position AccuracyRoboStream
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| RoboStreamModel size=32B2026.03 | 66 | 44.6 | 48 | 37.5 | 48.9 | |
| SoFarModel size=32B2026.03 | 60.4 | 41.9 | 45.8 | 33.3 | 45.3 | |
| SoFar2026.03 | 59.6 | 33.8 | 54.6 | 31.3 | 43.9 | |
| GPT-4o2026.03 | 49.4 | 28.4 | 44.2 | 25.8 | 36.2 | |
| RoboStreamModel size=8B2026.03 | 47.2 | 31.1 | 35.4 | 35.4 | 36.8 | |
| RoboPoint2026.03 | 43.8 | 30.8 | 33.8 | 25.8 | 33.5 | |
| SoFarModel size=8B2026.03 | 41.5 | 24.3 | 27.1 | 31.3 | 30.5 | |
| SpaceMantis2026.03 | 33.6 | 29.2 | 27.2 | 25 | 28.9 |