Multimodal AI — Glossary

AI that works across more than one kind of data — text, images, audio, and video — rather than text alone.

Multimodal AI handles more than one type of data — text, images, audio, video — in a single model or system. A multimodal model can describe a photo, read a chart, transcribe a meeting, or generate an image from a sentence. Text is just one of several inputs and outputs it understands.

This matters because real work rarely fits in plain text. A support request might arrive as a screenshot, a marketing brief might need fresh visuals, a recording might need a summary. With multimodal models you choose the right tool for each job — a large language model for copy and reasoning, an image model for creative — and combine them in one workflow.

We treat this as picking the best model for each task rather than forcing everything through one. In practice that's how systems like our AI-run Google Ads engine work: language models write and analyze, image models produce creative, and the right one runs at each step.