A mobile app team wants to generate a set of illustrations for their onboarding screens directly from short text descriptions. They plan to deploy a model in Microsoft Foundry and call it from a lightweight app. Which type of model should they deploy to produce the visual outputs from text prompts?