Describe an image
The other direction: a picture in, words out — enough to search with, or to write a prompt from.
Text and subject
Optical character recognition reads any text; a vision capability describes the subject. Both go through the shared capability catalogue, so this is not a second model to keep in step.
Availability is reported, not assumed
Like the image lane above, this needs a provider. The page states whether one is reachable rather than failing at the first press.