In the context of language models, a token is a unit of text — roughly a word, part of a word, or a punctuation mark — that the model processes as a single element.
Language models do not process text letter by letter or word by word in the way humans read. They break text into tokens, which are the basic units of input and output. A token might be a whole word like "data", a word fragment like "un" in "understanding", or a punctuation mark.
Token limits matter practically: most language models have a maximum number of tokens they can process in a single interaction (the context window). Very long documents or conversations may exceed this limit.