A token is a basic unit of text used by a Large Language Model (LLM).
It can be a whole word, part of a word, a character or punctuation.
On average, one token is about four characters or roughly 0.75 words. This is only an estimate and varies by text.
Tokens are how the model reads and writes language.
When you enter text, it is broken into tokens through a process called tokenisation.
The total token count includes both your input and the model’s response.
This total affects limits such as the context window. That determines how much text the model can ‘remember’ and use when generating a reply.
