A retailer wants a single model that can take a customer's photo of a damaged product plus a typed description of the problem, and return a written explanation of the likely cause along with a spoken summary for a call center agent to review. The input and output span text, images, and audio in one request.
What kind of model capability does this scenario require?