Gpt Tokenizer

This playground shows how text becomes tokens before a model ever sees it. You paste a prompt, choose a model, and the page tokenizes live, displaying the exact chunks the text is split into and a running token count, with an option to reveal every token rather than a truncated view. It is aimed at developers and prompt engineers who need to reason about cost and context before shipping. Two modes are offered: a single prompt input, and a chat prompt builder where you assemble a conversation message by message and see a tokenized conversation preview alongside the chat token total. Cost estimation sits beside it, with a breakdown per request that separates input, output, cached input and cached output, plus a separate estimate for the batch endpoint. A model insights panel reports the context window, maximum output, knowledge cutoff, input and output modalities, supported endpoints, supported features and the published price per million tokens for the selected model. The computed data can be exported, and a short primer explains what tokenization is for anyone meeting it for the first time. The tool is free. The structured data published on the page lists a price of zero, there is no pricing page on the domain and no paid tier of any kind, and the site links to the source repository on GitHub for the open source tokenizer library it is built on. The whole interface renders in the browser from a single client side bundle, so it stays usable as a quick reference without an account, an API key or a subscription.