Text-to-Video Generation on VideoPhy2 HARD
71.7PC ScoreProprio (Shybrid)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Proprio (Shybrid)Model=Hunyuan Video2026.05 | 71.7 | 28.9 | 25 | |
| Proprio (Sglobal)Model=Hunyuan Video2026.05 | 70.8 | 25 | 18.8 | |
| Proprio (Smotion)Model=Hunyuan Video2026.05 | 70.8 | 22.9 | 18.8 | |
| WMRewardModel=Hunyuan Video2026.05 | 70 | 27.2 | 21.7 | |
| Qwen 7BModel=Hunyuan Video2026.05 | 64.4 | 31.7 | 23.9 | |
| RandomModel=Hunyuan Video2026.05 | 61.1 | 29.4 | 20 | |
| Proprio (Shybrid)Model=Turbo Wan14B2026.05 | 58.3 | 33.3 | 23.9 | |
| Proprio (Smotion)Model=Turbo Wan14B2026.05 | 57.8 | 33.9 | 23.9 | |
| WMRewardModel=Turbo Wan14B2026.05 | 53.9 | 30 | 20 | |
| Proprio (Sglobal)Model=Turbo Wan14B2026.05 | 53.3 | 33.3 | 22.2 | |
| ProPhyBackbone=CogVideoX-5B2025.12 | 52.8 | 11.7 | 6.1 | |
| Proprio (Shybrid)Model=Wan2.2 5B2026.05 | 52.8 | 33.3 | 22.8 | |
| HunyuanVideo2025.12 | 52.2 | 7.2 | 5 | |
| VideoREPABackbone=CogVideoX-5B2025.12 | 52.2 | 7.8 | 5.6 | |
| WISABackbone=CogVideoX-5B2025.12 | 51.7 | 11.1 | 5 | |
| Proprio (Smotion)Model=Wan2.2 5B2026.05 | 51.7 | 34.4 | 23.3 | |
| CogVideoX-5BBackbone=CogVideoX-5B2025.12 | 51.1 | 9.6 | 5 | |
| Cosmos-Predict2.52025.12 | 50.6 | 8.3 | 5.6 | |
| Proprio (Sglobal)Model=Wan2.2 5B2026.05 | 50 | 30 | 18.9 | |
| ProPhyBackbone=Wan2.1-1.3B2025.12 | 48.9 | 12.2 | 7.2 | |
| WMRewardModel=Wan2.2 5B2026.05 | 47.8 | 33.9 | 20.6 | |
| RandomModel=Turbo Wan14B2026.05 | 47.8 | 32.2 | 23.3 | |
| Qwen 7BModel=Turbo Wan14B2026.05 | 46.1 | 32.8 | 22.2 | |
| Wan2.1-14BBackbone=Wan2.1-14B2025.12 | 45.6 | 8.3 | 3.9 | |
| RandomModel=Wan2.2 5B2026.05 | 44.4 | 31.7 | 19.4 | |
| Qwen 7BModel=Wan2.2 5B2026.05 | 42.2 | 35 | 18.9 | |
| Cosmos2025.12 | 40 | 5.1 | 4 | |
| Wan2.1-1.3BBackbone=Wan2.1-1.3B2025.12 | 36.7 | 11.7 | 5.6 |