The trusted token counter for GPT, Claude & Gemini. Exact GPT counts, accurate estimates, and live API cost — computed entirely in your browser.
| Model | Tokens | Input cost |
|---|
Exact = OpenAI's real tokenizer (tiktoken) running locally, byte-for-byte identical to the API. Estimated = Anthropic, Google and Meta don't publish their tokenizers, so those use a tuned approximation (typically within ~10–15%). Pricing is per 1M tokens, last verified June 2026. Always verify against official provider pages before billing: OpenAI pricing, Anthropic pricing, and Gemini pricing.
A token is the smallest chunk of text a language model reads — usually a short word or a part of a word. Models split text into tokens with a tokenizer, which is why token counts don't match word or character counts.
For English, roughly 1 word ≈ 1.3 tokens, or about 4 characters per token. Code, punctuation and non-English text use more tokens per word.
For GPT models the count is exact — it runs OpenAI's real tokenizer in your browser. Claude and Gemini don't publish their tokenizers, so those counts are close estimates and are labelled "Estimated".
No. All counting happens locally in your browser. Your text never leaves your device — there's no backend.
This token counter tells you exactly how many tokens your text uses before you send it to an AI model — so you can predict cost, stay inside the context window, and tighten your prompts. GPT counts are exact: the page runs OpenAI's real tiktoken tokenizer (o200k_base) directly in your browser, so the number is byte-for-byte identical to what the API bills you. Claude and Gemini don't publish their tokenizers, so those models show a carefully tuned estimate, clearly labelled "Estimated".
Everything runs locally. Your text is never uploaded, logged or stored — there is no backend, which makes this safe for confidential prompts, proprietary code and private documents.
For English text the handy rule of thumb is 1 token ≈ 4 characters ≈ 0.75 words — so 1,000 tokens is roughly 750 words. Code, punctuation and non-English text use more tokens per word. New here? Start with what is a token in AI, then see tokens vs words vs characters and how many tokens is my text.