A token is a fragment of a word: in French, about 0.75 of a word or four characters. The model splits your text into tokens, reads some (input) and then produces others (output), and every token generated uses energy. It is therefore the volume of tokens processed, especially on the output side, rather than the number of queries, that determines the energy a ChatGPT query uses.
Read the source article: How much energy does a ChatGPT query use? About 0.3 Wh in 2026