Video spatial & 4D reasoning on OSI-Bench 2025
47.9AccuracySpatialClaw
Evaluation Results
| Method | Links | |
|---|---|---|
| SpatialClawModel=Qwen3.6-27B2026.06 | 47.9 | |
| SpatialClawModel=Qwen3.5-397B-A17B2026.06 | 45.6 | |
| SpatialClawModel=Qwen3.6-35B-A3B2026.06 | 41.9 | |
| SpatialClawModel=Gemma4-31B2026.06 | 41.9 | |
| SpatialClawModel=Qwen3.5-122B-A10B2026.06 | 41.4 | |
| SpatialClawModel=Gemma4-26B-A4B2026.06 | 40 | |
| No-toolModel=Qwen3.5-397B-A17B2026.06 | 37.3 | |
| No-toolModel=Qwen3.5-122B-A10B2026.06 | 37.1 | |
| No-toolModel=Qwen3.6-27B2026.06 | 36.8 | |
| No-toolModel=Gemma4-31B2026.06 | 35.6 | |
| No-toolModel=Qwen3.6-35B-A3B2026.06 | 35.5 | |
| No-toolModel=Gemma4-26B-A4B2026.06 | 31.1 |