I tested out a buzzy new text-to-video AI model from China