This article presents the Vision Bridge Transformer (ViBT), a model designed for efficient image and video translation. It features two parameter variants, optimized training methods, and faster inference by simplifying token usage. The authors also outline specific tasks for image and video processing, with training code forthcoming.