Spatial VQA on EmbSpatial
80.71AccuracyHiSpatial-3B
Evaluation Results
| Method | Links | |
|---|---|---|
| HiSpatial-3BInput=RGB-XYZ2026.03 | 80.71 | |
| HiSpatial-3B-RGBInput=RGB2026.03 | 79.78 | |
| Qwen-3-VL-8BInput=RGB2026.03 | 78.5 | |
| Gemini-2.5-ProInput=RGB2026.03 | 76.67 | |
| Qwen-3-VL-30B-A3BInput=RGB2026.03 | 76.4 | |
| RoboRefer-8B-SFTInput=RGB-D2026.03 | 72.53 | |
| InternVL3.5-14BInput=RGB2026.03 | 71.62 | |
| InternVL3.5-8BInput=RGB2026.03 | 69.81 | |
| GPT-4oInput=RGB2026.03 | 63.38 | |
| SpatialRGPT-8BInput=RGB-D2026.03 | 59.62 | |
| SpatialBot-3BInput=RGB-D2026.03 | 50.66 | |
| SpaceLLaVA-13BInput=RGB2026.03 | 49.4 | |
| Claude-3.7-SonnetInput=RGB2026.03 | 33.33 | |
| PaliGemma2-3BInput=RGB2026.03 | 28.32 |