Glossary
Multimodal model — Glossary
A model that accepts or generates more than one data type, such as text and images.
Full definition
Multimodal systems connect encoders for different media into a shared or coordinated representation space. Examples include vision-language chat models and text-to-image generators.
Evaluate each modality’s errors: visual hallucinations, OCR mistakes, and audio noise can hide behind fluent text answers. Show users the evidence used whenever possible.