On October 27th, the Meituan LongCat team proudly announced the official release and open-sourcing of the LongCat-Video video generation model. This cutting-edge model has attained state-of-the-art (SOTA) performance in the open-source realm for both text-to-video and image-to-video tasks. By undergoing pre-training on native video continuation tasks, LongCat-Video is capable of generating coherent minute-long videos, all the while maintaining cross-frame temporal consistency and ensuring that the physical motions depicted are plausible. The Meituan LongCat team emphasized that this release represents a significant milestone in their exploration of the "world model." Looking ahead, the LongCat model will harness its precise reconstruction capabilities of real-world operational states to drive applications in highly interactive business scenarios, including autonomous driving and embodied AI.