we have no plans for image or other modalities. I'd like to keep it just text-to-text so it can be as sharp of a tool as possible