A multimodal model that bridges vision and language by combining image understanding with text generation. It takes in both images and text prompts, then produces descriptive or question-answering responses grounded in visual content. Works well for visual question answering and image captioning tasks, though its context window is minimal and it's not suited for long-form reasoning.