Foundation Model, because multimodal models aren't just Language