Last reviewed
Correct answer: D. It can handle multiple types of input, like text and images
Explanation
Multimodal describes what the model accepts on the way in, and the consequence matters more than the definition. Images, audio, video, PDFs and other documents ride in the same request as the text prompt, as entries in one input list, not through some separate vision endpoint with its own key. Small payloads go inline as base64 or by URL; anything pushing the request past 100 MB goes through the Files API first, with a 50 MB ceiling on PDFs.
That is what makes captioning, classification and visual question answering possible without training a specialist model, and it is why a single call can hand over a scanned invoice, ask which line items were taxed, and get a written answer.
Multilingual is a different property: a text-only model can answer in twenty languages and still be single-modality, so option B confuses output language with input type. Accounts are unrelated — one key on one project covers every modality. And audio is one accepted input among several rather than the only one, which is where option D goes wrong.
Sources
“Gemini can handle various types of input data, including text, images, and audio, at the same time.”
“Gemini models are built to be multimodal from the ground up”
“Transform text, images, video and audio into rich interactive user interfaces.”
Practise 2 questions on this topic
Take Gemini Basics — Timed Test (2 questions) — scored instantly, explanation for every question, no login.