Developer / Tokens
Token Counter: Count Tokens for OpenAI Models
Count the tokens in any text exactly as OpenAI's models do, with the o200k_base tokenizer used by GPT-4o and newer and cl100k_base used by GPT-4 and GPT-3.5 Turbo, alongside characters, words, and characters per token, and see each token highlighted, all without sending your text anywhere.
Token Counter: Count Tokens for OpenAI Models: The text is encoded in your browser with js-tiktoken and OpenAI's published byte-pair vocabularies, the same ones the models use, so counts match exactly: "tiktoken is great!" is six tokens in cl100k_base. Each token is shown as a coloured piece, so you can see that common words are one token while rare words, numbers, and other scripts split into several. Runs 100% locally in your browser with zero server file uploads.
- Category
- Developer tools
- Runs
- In your browser
- Cost
- Free · no sign-up
- Availability
- Ready to use
Runs entirely in your browser
Counts are exact for OpenAI models that use these encodings, computed in your browser with js-tiktoken (MIT) and OpenAI's published vocabularies; your text is not sent anywhere. Other companies' models use their own tokenizers, so treat these counts as an estimate for them: in English, a token is often about four characters.
Tokens and cost
Model prices are per million input and output tokens, so token counts let you estimate a request's cost; check the provider's current price list.
Rule of thumb
In English, 100 tokens is about 75 words, or four characters a token; other languages and code usually need more tokens per word.
How to use it
- Paste your prompt or text.
- Choose the tokenizer for your model.
- Read the token count, and look at how the text was split.
Privacy & limitations
Your text is tokenized in your browser and never uploaded.
Related tools
Frequently asked questions
Does this count tokens for Claude, Gemini, or Llama?
Not exactly: each company uses its own tokenizer, so counts for other models can differ noticeably. Use the count here as a rough estimate for them.
Why does Chinese or Japanese use more tokens?
Vocabularies learn the most common pieces of their training text. o200k_base has more non-English pieces than cl100k_base, so the same Japanese text usually takes fewer tokens with it.
Do spaces and line breaks count?
Yes. A space is usually joined to the word after it, and line breaks are tokens too.
Free tool · runs in your browser · no account required