Multimodal
Also known as: multimodal model
Multimodal means the model works across several modalities — text, images, audio, video — not just one.
Examples:
- Text → text: a plain chatbot.
- Text + image → text: “describe this photo.”
- Text → image: an image generator.
- Audio → text: a transcription model.
Modern frontier models are increasingly natively multimodal.
Where it's heading: the frontier is converging on models that take any combination of inputs and produce any combination of outputs. Multimodal is the direction of travel.