Cover Image

Saved in:
Bibliographic Details
Main Authors: ai, Sand., Teng, Hansi, Jia, Hongyu, Sun, Lei, Li, Lingzhi, Li, Maolin, Tang, Mingqiu, Han, Shuai, Zhang, Tianning, Zhang, W. Q., Luo, Weifeng, Kang, Xiaoyang, Sun, Yuchen, Cao, Yue, Huang, Yunpeng, Lin, Yutong, Fang, Yuxin, Tao, Zewei, Zhang, Zheng, Wang, Zhongshu, Liu, Zixun, Shi, Dai, Su, Guoli, Sun, Hanwen, Pan, Hong, Wang, Jie, Sheng, Jiexin, Cui, Min, Hu, Min, Yan, Ming, Yin, Shucheng, Zhang, Siran, Liu, Tingting, Yin, Xianping, Yang, Xiaoyu, Song, Xin, Hu, Xuan, Zhang, Yankai, Li, Yuqiao
Format: Preprint
Published: 2025
Subjects:
Computer Vision and Pattern Recognition
Artificial Intelligence
Online Access:https://arxiv.org/abs/2505.13211
Tags: Add Tag
No Tags, Be the first to tag this record!
Table of Contents:
  • We present MAGI-1, a world model that generates videos by autoregressively predicting a sequence of video chunks, defined as fixed-length segments of consecutive frames. Trained to denoise per-chunk noise that increases monotonically over time, MAGI-1 enables causal temporal modeling and naturally supports streaming generation. It achieves strong performance on image-to-video (I2V) tasks conditioned on text instructions, providing high temporal consistency and scalability, which are made possible by several algorithmic innovations and a dedicated infrastructure stack. MAGI-1 facilitates controllable generation via chunk-wise prompting and supports real-time, memory-efficient deployment by maintaining constant peak inference cost, regardless of video length. The largest variant of MAGI-1 comprises 24 billion parameters and supports context lengths of up to 4 million tokens, demonstrating the scalability and robustness of our approach. The code and models are available at https://github.com/SandAI-org/MAGI-1 and https://github.com/SandAI-org/MagiAttention. The product can be accessed at https://sand.ai.

Similar Items