Meet Our New AI Token Counter

A short prompt can become a much larger request once you add instructions, conversation history and documents. Our new AI token counter and context inspector helps you see where that space goes before sending anything to a model. It is free, runs locally in your browser, and is available in English and French. It does not use AI to analyze your text.

See what a rewrite changes

Compare an original prompt with a shorter draft, or look at the difference between formatted and compact JSON. The counter shows the token split as well as the total, so you can see how punctuation, spacing and different languages affect the result. Ready-made examples give you something to explore without pasting your own content.

Choose the encoding that matches your model: the tool supports o200k_base and cl100k_base. A smaller count is useful information, but it does not prove that a rewritten prompt will work better.

Leave room for the answer

When a request includes several pieces of context, assemble them and check how much room remains after reserving space for an answer. You can also try an editable USD cost scenario before repeating the request hundreds of times. These are text-based estimates: API formatting and other provider charges still need checking.

If you are preparing the request body too, our guide to JSON, form and multipart requests explains how those formats differ.

Try the AI token counter with a prompt or one of the built-in examples.