Interactive Video World Model Evaluation on RealWM120K VBench (val)
11Latency (s)Matrix-Game 2.0
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Matrix-Game 2.0Batch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 11 | 80.82 | 0.9457 | 0.9814 | 0.8476 | 0.899 | 0.4971 | 0.6784 | |
| MagicWorldBatch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 15 | 85.47 | 0.9752 | 0.9921 | 0.9627 | 0.9408 | 0.5394 | 0.7182 | |
| YUME 1.5Batch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 19 | 83.34 | 0.9502 | 0.9821 | 0.9115 | 0.9221 | 0.527 | 0.7077 | |
| Wan2.1-CameraBatch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 39 | 81.72 | 0.9586 | 0.9801 | 0.8778 | 0.9173 | 0.5018 | 0.6674 | |
| Wan2.2-CameraBatch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 94 | 79.35 | 0.9573 | 0.9846 | 0.8508 | 0.8982 | 0.4861 | 0.5837 | |
| HY-World 1.5Batch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 163 | 84.52 | 0.9719 | 0.9918 | 0.9409 | 0.9348 | 0.5303 | 0.7014 | |
| ViewCrafterBatch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 302 | 78.07 | 0.9569 | 0.979 | 0.8188 | 0.8748 | 0.5001 | 0.5543 | |
| YUMEBatch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 1,918 | 83.14 | 0.9491 | 0.9865 | 0.9098 | 0.9264 | 0.5239 | 0.6926 | |
| LingBot-World-BaseBatch size=1, Video dimensions=480×832×81, GPU=NVIDIA H202025.11 | 1,920 | 83.64 | 0.9653 | 0.9755 | 0.9423 | 0.9261 | 0.5406 | 0.6687 |