Multimodal AI
Vision-language model
VLM
A model that jointly understands visual information and language.
Example
A vision-language model describes an image and answers questions about its contents.
Why it matters
It lets people ask ordinary-language questions about visual information.
You'll hear it
“Which part of the diagram supports that answer?”
Can you control it?
Sometimes
Sometimes. Some products or workflows let you influence a vision-language model, while others keep it fixed or manage it automatically. Check the settings and documentation for the tool you are using.
Common questions
- Can it answer with details that are not visible?
- Yes. It may infer or invent information rather than stick to the image.
- Can written instructions conflict with a picture?
- Yes. The system may need clarification when the two suggest different tasks or facts.
- Does it read every small label reliably?
- No. Resolution, layout and text size can limit what it can interpret.