Visual Spatial Inference on VSI-Bench Tiny (video-input)
69Object Count ScoreReasmory (Gemini-3-flash)
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Reasmory (Gemini-3-flash)Base Model=Gemini-3-flash, Method Category=Other Test-time Scaling Methods2026.05 | 69 | 50.4 | 68.4 | 62 | 71.4 | 65 | 62 | 72 | 65 | |
| Spatial-MLLM-4BMethod Category=Spatial Models2026.05 | 67.2 | 31 | 56.6 | 43.6 | 32 | 52 | 36 | 44 | 45.2 | |
| Reasmory (GPT-5-mini)Base Model=GPT-5-mini, Method Category=Other Test-time Scaling Methods2026.05 | 60.2 | 41.9 | 73.1 | 50.2 | 52 | 55.6 | 42 | 71.4 | 55.8 | |
| SpatialLadder-3BMethod Category=Spatial Models2026.05 | 60 | 31 | 54.2 | 47.6 | 30 | 44 | 30 | 40 | 42.1 | |
| GPT-5-miniBase Model=GPT-5-mini2026.05 | 49 | 31.6 | 73 | 41.2 | 42 | 48 | 46 | 66 | 49.6 | |
| Mindjourney (GPT-5-mini)Base Model=GPT-5-mini, Method Category=Other Test-time Scaling Methods2026.05 | 47 | 25 | 64.5 | 23 | 20 | 50 | 30 | 50 | 38.7 | |
| Gemini-3-flash (16 frames)Base Model=Gemini-3-flash, Input=16 frames2026.05 | 46.4 | 15.2 | 68.6 | 48.4 | 56 | 58 | 48 | 72 | 51.6 | |
| Gemini-3-flash (video)Base Model=Gemini-3-flash, Input=video2026.05 | 45.9 | 21.8 | 75 | 39.2 | 58 | 62 | 40 | 88 | 53.7 | |
| Think3D (Gemini-3-flash)Base Model=Gemini-3-flash, Method Category=Other Test-time Scaling Methods2026.05 | 45.8 | 10.2 | 62.8 | 46.4 | 54 | 52 | 58 | 52 | 47.3 | |
| Think3D (GPT-5-mini)Base Model=GPT-5-mini, Method Category=Other Test-time Scaling Methods2026.05 | 42.4 | 25.2 | 63.8 | 41.2 | 34 | 48 | 31 | 71.7 | 44.7 | |
| SpaceOM-3BMethod Category=Spatial Models2026.05 | 27.3 | 10.2 | 17.1 | 21.7 | 10.2 | 25 | 22.4 | 31.9 | 20.7 | |
| Mindjourney (Gemini-3-flash)Base Model=Gemini-3-flash, Method Category=Other Test-time Scaling Methods2026.05 | 25.5 | 36 | 59.5 | 56.5 | 35 | 50 | 65 | 80 | 50.9 |