TMTPOST -- The experimental video generation model "VideoWorld" was jointly developed by the Doubao Large Model Team, Beijing Jiaotong University, and the University of Science and Technology of China. Unlike mainstream multimodal models like Sora, DALL-E, and Midjourney, VideoWorld is the first in the industry to achieve world understanding purely through visual cognition, without relying on language models. The project's code and model have now been made open-source.