Token

A token is the unit a large language model uses to process text. The model does not read whole pages or even whole words. It splits text into tokens, which can be a short word, a fragment of a longer word, a number or a punctuation mark. As a rough rule of thumb for English, a token is about four characters, so a typical paragraph runs to a few dozen tokens.

Why do tokens matter for PR and communications teams?

Tokens set the limits on what an AI engine can take in at once. The number of tokens a model can hold in a single exchange is its context window, so a long document, a long prompt and the model's own reply all draw from the same budget. Vendors also price API use per token.

Tokens also explain why structure matters for AI visibility. Engines retrieve passages instead of whole pages, so the first few hundred tokens of a page carry disproportionate weight. A page that states its answer in the opening paragraph gives a retrieval system something clean to lift. See Retrieval Chunking and Answer Formatting for how that plays out.

How do tokens relate to other terms?

A large language model predicts the next token again and again to produce an answer. The knowledge cutoff marks the point where the tokens in its training data stop. Anything after that has to arrive through retrieval, which is what grounding does.