Starchild-1: The First Real-Time Multimodal World Model

I'm a little confused on what "simulating in real time" means based on the videos and description.

Because the videos are generated and pre-recorded on the website first?

Context: not well versed in machine learning