MAGI-1: Autoregressive Video Generation at Scale

TL;DR

MAGI-1 is a scalable autoregressive video generation model that produces high-quality, temporally consistent videos from text prompts, supporting real-time streaming and controllable generation with a massive 24-billion-parameter architecture.

Contribution

This work introduces MAGI-1, a novel large-scale autoregressive video model that enables causal, streaming, and controllable video generation with unprecedented scalability and efficiency.

Findings

Achieves high temporal consistency in generated videos.

Supports context lengths up to 4 million tokens.

Maintains constant peak inference cost regardless of video length.

Abstract

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.