A multimodal model that takes both images and text as input, producing text responses. Built on the BLIP and Flan-T5 XL architecture, it handles visual question answering and image-grounded conversations. As a smaller open-weight model, it trades raw capability for accessibility and ease of deployment.