LLM

model

An LLM works with text alone: you give it words and it gives words back. A VLM is the same idea with an image added to the prompt.

For engineers: a text-only generative model; the vision variant (VLM) extends it with an image encoder.