Vision-language model (VLM)
A language model that also reads images: it names a defect, reads a label, or reads a field on a scanned page.
Category
How a model is chosen and adapted
Related terms
A vision-language model (VLM) is a language model that also reads images, so it can say in words what a picture or a scanned page shows: the type of a defect, the text on a label, a field on a form.
What it is used for
Explaining and classifying what another step found, reading labels, and extracting fields from scanned documents. It is one kind of small model like the others: tested on graded cases first, fine-tuned on the customer's own images and terms, and run in the customer's environment.
What it does not decide
On a production line, pass or fail is decided by a model trained on good parts, inside thresholds the quality manager sets, not by the vision-language model. Video is not offered. The visual inspection use case shows the split: the anomaly model makes the gate call, and the vision-language model explains and classifies, never deciding pass or fail.
Want these terms worked through on one of your own decisions? Start with a Decision Blueprint