Token (AI Language Model)
A token is the smallest unit of text a language model works with – often part of a word, a whole word or a punctuation mark – and both the context window and the cost of API usage are measured in tokens rather than in characters or words.
In practice
This explains why very long prompts or documents can hit model limits, and why the answer length of a model is capped. German and other non-English languages often need more tokens per word than English because of compound words and umlauts, which matters when calculating API costs. If you work with language models, keep an eye on the token count of your content.