AI Glossary Term
What does “Vision-Language Model (VLM)” mean?
Definition
A multimodal model that jointly processes visual and textual information.
How Vision-Language Model (VLM) works
Example
You upload a photo of a chart and ask, 'Which quarter had the highest sales?' The model reads the image and answers in text. It may still misread small numbers, so it is worth checking.
Why it matters
VLMs bring image understanding into conversational tools. They can describe things that are not in the image if they guess wrongly.
Common questions
Common uses?
Describing images, reading charts and documents, answering questions about photos, and assisting people with low vision.
VLM vs CLIP?
CLIP matches images and text in a shared space. A VLM can also generate text answers about an image.
Can they make mistakes?
Yes. They can miss details, misread text, or state things not present in the image.