Spatial VQA on RoboSpatial
86.18AccuracyHiSpatial-3B
Evaluation results
| Method | Links | |
|---|---|---|
| HiSpatial-3BInput=RGB-XYZ2026.03 | 86.18 | |
| RoboRefer-8B-SFTInput=RGB-D2026.03 | 84.55 | |
| HiSpatial-3B-RGBInput=RGB2026.03 | 83.74 | |
| Qwen-3-VL-30B-A3BInput=RGB2026.03 | 83.73 | |
| Qwen-3-VL-8BInput=RGB2026.03 | 82.11 | |
| InternVL3.5-8BInput=RGB2026.03 | 78.86 | |
| InternVL3.5-14BInput=RGB2026.03 | 78.86 | |
| Gemini-2.5-ProInput=RGB2026.03 | 77.24 | |
| GPT-4oInput=RGB2026.03 | 77.2 | |
| SpatialBot-3BInput=RGB-D2026.03 | 72.36 | |
| PaliGemma2-3BInput=RGB2026.03 | 71.54 | |
| SpatialRGPT-8BInput=RGB-D2026.03 | 66.67 | |
| SpaceLLaVA-13BInput=RGB2026.03 | 61 | |
| Claude-3.7-SonnetInput=RGB2026.03 | 60.73 |