A compact multimodal model that punches at its weight class — handling both text and image inputs within a 1B parameter footprint. It processes visual and language information together, making it accessible for resource-constrained environments. The small size means capability trade-offs compared to larger siblings, but it remains deployable on modest hardware.