MAGI 1 - AUTOREGRESSIVE VIDEO GENERATION AT SCALE (HUGGING FACE HUB)

Summary from the TLDR AI newsletter

The MAGI 1 model is a new autoregressive video generation model that can generate long form and consistent videos. It performs on par with Wan video generation while being slightly below some closed models.