Skip to main content
The inference API supports six output modalities. Each maps to an SDK method. You pick a model that supports the modality you need (use the catalog to search), then call the matching method.

All modalities

Text

Image

Embeddings

Vision

Send an image alongside text by using content parts in the message: