Human-centric Spatial Reasoning on H*Bench transfer evaluation
70.1Overall ScorePanoWorld
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PanoWorldSetting=ERP panorama-based, Training=H* SFT2026.05 | 70.1 | 73.1 | 64.2 | 74.1 | 85.5 | |
| PanoWorldSetting=Perspective-based, Mode=zero-shot2026.05 | 56.1 | 61.8 | 47.5 | — | — | |
| Gemini-2.5-ProSetting=ERP panorama-based2026.05 | 46.9 | 55.3 | 34.3 | 52.5 | 71.6 | |
| Qwen3.5-9BSetting=ERP panorama-based, Prompting=Visual prompt2026.05 | 40.4 | 46 | 32 | 43.5 | 52 | |
| Qwen3.5-9BSetting=ERP panorama-based, Prompting=Text prompt2026.05 | 38.5 | 43.3 | 31.2 | 40 | 49.5 | |
| HVS-3B*Setting=Perspective-based2026.05 | 38.4 | 47.3 | 24.9 | — | — | |
| Gemini-2.5-ProSetting=Perspective-based2026.05 | 32.3 | 31.9 | 33 | — | — | |
| GPT-4oSetting=ERP panorama-based2026.05 | 30.1 | 39.1 | 17.1 | 38.5 | 64.2 | |
| GPT-4oSetting=Perspective-based2026.05 | 21.3 | 19.7 | 23.6 | — | — | |
| Qwen3.5-9BSetting=ERP panorama-based, Mode=zero-shot2026.05 | 19.4 | 26.2 | 9.3 | 23.5 | 46.5 | |
| Qwen3-VL-8BSetting=Perspective-based2026.05 | 19.1 | 23.6 | 12.2 | — | — | |
| Qwen3.5-9BSetting=Perspective-based2026.05 | 18.9 | 21.8 | 14.5 | — | — | |
| Qwen3.5Setting=ERP panorama-based, Training=H* SFT2026.05 | 17.8 | 11.1 | 27.7 | 25.3 | 42.5 | |
| Qwen3-VL-4BSetting=Perspective-based2026.05 | 17.5 | 19.5 | 14.4 | — | — | |
| InternVL3.5-8BSetting=ERP panorama-based2026.05 | 14.9 | 18 | 10.25 | 19.2 | 38.5 | |
| Qwen3-VL-8BSetting=ERP panorama-based2026.05 | 13.1 | 15 | 10.3 | 16.8 | 39.2 | |
| Qwen3-VL-4BSetting=ERP panorama-based2026.05 | 12.8 | 14.3 | 10.5 | 15.6 | 40 | |
| Gemma-3-12BSetting=Perspective-based2026.05 | 11.9 | 10.2 | 14.5 | — | — | |
| InternVL3.5-4BSetting=ERP panorama-based2026.05 | 11.6 | 12.8 | 9.75 | 22.1 | 41.5 | |
| Qwen2.5-VL-3BSetting=Perspective-based2026.05 | 11.4 | 14.8 | 6.4 | — | — | |
| Gemma-3-12BSetting=ERP panorama-based2026.05 | 9.6 | 10.9 | 7.5 | 14 | 40.8 | |
| Qwen2.5-VL-7BSetting=Perspective-based2026.05 | 9.3 | 11.3 | 6.3 | — | — | |
| Qwen2.5-VL-7BSetting=ERP panorama-based2026.05 | 9.1 | 9.6 | 8.5 | 11.9 | 32.3 | |
| InternVL3.5-8BSetting=Perspective-based2026.05 | 6.7 | 6.4 | 7.2 | — | — | |
| Kimi-VL-A3BSetting=Perspective-based2026.05 | 4.6 | 4.92 | 4.32 | — | — | |
| InternVL3.5-4BSetting=Perspective-based2026.05 | 3.8 | 3.2 | 4.8 | — | — |