A mid-sized vision-language model that can look at an image and talk about it in coherent, detailed prose. It handles everyday visual questions — describing scenes, reading text in images, reasoning about diagrams — with reasonable fluency, though it can stumble on fine-grained spatial details or complex multi-step visual reasoning. Built on Vicuna's conversational backbone, it tends to be chatty and explanatory rather than terse.