A token is one discrete symbol produced by a tokenizer for model input or output.
Tokens can represent whole words, word fragments, punctuation, bytes, or other units depending on the tokenizer.
Token count affects context usage, inference cost, and output limits. A token is not the same as a character or a word.
Related knowledge
Context engineering for LLM systemsDeliberately select and structure the information a model receives instead of treating the context window as an unbounded memory store.