Visual Grounding on 3D FORCE REF 1.0 (test)
96.79REF ScoreSATURN
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SATURNOracle 3D=true2026.06 | 96.79 | — | 95.94 | |
| SATURNBackbone=Qwen3-VL-8B-Instruct2026.06 | 89.56 | — | 88.85 | |
| Qwen3.5-35BParameters=35B2026.06 | 67.53 | 99.76 | 64.33 | |
| Qwen3.5-9BParameters=9B2026.06 | 62.84 | 98.75 | 67.08 | |
| Gemini-3.1-Pro2026.06 | 47.56 | 91.19 | 57.41 | |
| InternVL3.5-38BParameters=38B2026.06 | 19.68 | 99.81 | 33.35 | |
| InternVL3.5-8BParameters=8B2026.06 | 13.03 | 99.04 | 27.33 | |
| GCAtype=tool-augmented2026.06 | 9.5 | — | 24.24 | |
| GPT-5.12026.06 | 3.26 | 99.43 | 27.46 | |
| Qwen3-VL-8BParameters=8B2026.06 | 2.11 | 99.28 | 24.15 | |
| Random choice2026.06 | 1.53 | — | 18.8 | |
| Molmo2-8BParameters=8B2026.06 | 1.44 | 98.61 | 20.32 | |
| Qwen3-VL-235BParameters=235B2026.06 | 1.34 | 99.81 | 26.87 | |
| SpaceQwen-3BParameters=3B2026.06 | 0.86 | 72.22 | 13.96 | |
| SpaceThinker-3BParameters=3B2026.06 | 0.67 | 97.13 | 17.45 | |
| SpaceOm-3BParameters=3B2026.06 | 0.38 | 95.35 | 17.79 | |
| Cosmos-Reason1-7BParameters=7B2026.06 | 0.24 | 78.74 | 13.25 |