Spatial Reasoning QA on SAT-Real
88.7Average AccuracyGemini-3.0 + 3D Belief
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Gemini-3.0 + 3D BeliefModels=Gemini-3.0 + 3D Belief2026.05 | 88.7 | 100 | 73.9 | 97.3 | 94.1 | 75.8 | |
| Gemini-3.0 + SWMModels=Gemini-3.0 + SWM2026.05 | 86.7 | 95 | 70 | 94.1 | 89.7 | 81.3 | |
| Gemini-3.0 + SVCModels=Gemini-3.0 + SVC2026.05 | 86 | 95.7 | 78.3 | 86.5 | 94.1 | 75.8 | |
| Gemini-3.0Models=Gemini-3.02026.05 | 85.3 | 95.7 | 78.3 | 83.8 | 85.3 | 84.8 |