Video Reasoning on Sokoban (test)
34PrecisionVideoRLVR
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| VideoRLVRModel Category=RL Model, Training Epochs=1 (from SFT Epoch 5)2026.05 | 34 | 12.5 | 29.4 | 6.1 | |
| SFT Epoch 5Model Category=SFT Models, Training Epochs=52026.05 | 33.6 | 11.9 | 17.6 | 2.9 | |
| SFT Epoch 10Model Category=SFT Models, Training Epochs=102026.05 | 32.8 | 11.6 | 17.1 | 2.7 | |
| Veo 3.1Model Category=Proprietary2026.05 | 22.2 | 6 | 9.4 | 0 | |
| VBVR-Wan2.2Model Category=SFT Models2026.05 | 16.2 | 1.7 | 3.1 | 0 | |
| Sora 2Model Category=Proprietary2026.05 | 8.5 | 4.8 | 5.4 | 0 | |
| HunyuanVideoModel Category=Open-Source2026.05 | 8.2 | 2.7 | 3.2 | 0 | |
| Wan-R1Model Category=SFT Models2026.05 | 7.7 | 2.1 | 3.3 | 0 | |
| Kling V3Model Category=Proprietary2026.05 | 5.7 | 2.7 | 3.7 | 0 | |
| Wan2.2-TI2V-5BModel Category=Open-Source2026.05 | 4.1 | 0.7 | 1 | 0 | |
| CogVideoX1.5Model Category=Open-Source2026.05 | 3.2 | 0.3 | 0.5 | 0 |