A token is the basic unit of text a language model reads and generates. Tokens are typically word fragments rather than whole words — a common English word is often one token, while longer or unusual words split into several.
Models convert text into tokens before processing, and their context window — the amount of text they can consider at once — is measured in tokens. Pricing for commercial AI APIs is almost always per token, covering both input and output.
Tokens matter practically rather than conceptually. Context limits determine how much content can be supplied to a model at once, which constrains how retrieval systems chunk documents. Token pricing also determines the cost of any AI-assisted workflow at scale.
When supplying content to a model, prioritise the material that actually answers the question rather than pasting entire pages, since context is finite and costly. For content intended to be retrieved, keep sections self-contained so a useful chunk fits comfortably within a retrieval window.
Want to learn more? Explore our SEO Glossary or check out our expert guides on SEO strategies.