Hi! Thanks for the reply. I would just want to know whether image as an input modality is supported for the on device foundational model. Thanks