A native multimodal model with video, image, text and file input for visual coding and document work.