Tokens
The small chunks of text, roughly three-quarters of an English word each, that language models read and write.
Also: token
Language models do not see letters or whole words but tokens: common words, parts of words or punctuation, each mapped to a number. In English a token averages about four characters, so 1,000 tokens are roughly 750 words. Many tokenisers split Indian and other non-Latin scripts into more tokens per word, which makes the same text costlier to process. Tokens are the unit of AI work: model APIs are priced per million tokens, context windows are measured in tokens, and providers report output as tokens per second.