Pusa v1.0: Surpassing Wan-I2V with $500 Training Cost by Vectorized Timestep Adaptation
Published in ICLR 2026, 2026
Recommended citation: Yaofang Liu, Yumeng Ren, Aitor Artola, Yuxuan Hu, Xiaodong Cun, Xiaotong Zhao, Alan Zhao, Raymond H. Chan, Suiyun Zhang, Rui Liu, Dandan Tu, Jean-Michel Morel (2026). "Pusa v1.0: Surpassing Wan-I2V with $500 Training Cost by Vectorized Timestep Adaptation." ICLR 2026. https://arxiv.org/abs/2507.16116
We present Pusa v1.0, an Image-to-Video model that surpasses Wan-I2V with only $500 training cost using vectorized timestep adaptation.
| Paper | Project Page | Code |
Recommended citation: Yaofang Liu, Yumeng Ren, Aitor Artola, Yuxuan Hu, Xiaodong Cun, Xiaotong Zhao, Alan Zhao, Raymond H. Chan, Suiyun Zhang, Rui Liu, Dandan Tu, Jean-Michel Morel (2026). “Pusa v1.0: Surpassing Wan-I2V with $500 Training Cost by Vectorized Timestep Adaptation.” ICLR 2026.