Skip to main content

    Guides

    The counting rules behind the tools, and the assumptions behind every estimate. Organised by the question you are asking, not by keyword.

    Counting rules

    Word counter →

    What the engine treats as a word, a character, a letter and a sentence — and why tools disagree.

    Converting between words, characters, pages and minutes. Every answer here is an estimate with a stated assumption.

    Text structure

    Sentence counter →

    Sentence length, paragraph shape and trimming text that has to fit a limit.

    The long-form background on how text measurement works end to end.

    Prompts and tokens

    Tokens are not words. OpenAI models are tokenised exactly in your browser; other models are shown as an estimated range with the assumption stated.

    Prompt workspace

    Measured token counts for cl100k_base and o200k_base, estimated ranges for the rest, and a token budget you set yourself.