Language models do not process text letter by letter or as whole sentences; they break it into tokens. A token is roughly four characters of English, so a common word may be one token while a longer word splits into several.
Tokens matter for two practical reasons. First, a model can only handle so many tokens at once — its context window — which limits how much text it can consider. Second, paid AI services usually charge per token, for both your input and the model's output.
For example, the phrase AI is useful is about four tokens, and a 500-word document is roughly 650 to 700 tokens. Understanding tokens helps explain why very long documents can hit limits or cost more to process.