What is a Tokens (AI)?

A model does not see raw text; it sees a sequence of token IDs produced by tokenization, and it generates its response one token at a time, predicting the next token from everything before it. A word like automation might be one token, while a rarer or compound word might split into two or three. As a rough rule of thumb, one token is roughly three-quarters of an English word, so a 1,000-word document is typically around 1,300 tokens. Tokens matter practically in two ways: every model has a maximum context window measured in tokens, where prompt plus conversation history plus output must fit inside it, and API usage is billed per token, usually with input and output priced differently. Long conversations, large attached documents, or verbose system prompts all consume the same token budget, which is why summarization and context trimming become necessary as an agent's session grows.