MeridiaanCharacters and Token counter
Editor
How to use it.
- 1
Paste or type.
The counts update as you type. There is nothing to submit.
- 2
Read the numbers.
Characters, words, sentences, paragraphs, lines and bytes are exact. Tokens are an estimate.
- 3
Size your prompt.
Use the token estimate to see whether a document fits in your Agent’s context window, and what to trim when it does not.
What is exact and what is estimated.
Exact counts.
Characters, words, sentences, paragraphs, lines and bytes are exact. A letter with an accent or an emoji counts as one character, and bytes are counted in UTF-8, where é takes 2 bytes and 😀 takes 4. Reading time uses 238 words per minute, the average for silent reading of non-fiction in Brysbaert’s 2019 review of reading studies.
How tokens are estimated.
Models read tokens, not characters, and each model splits text its own way. The estimate starts from the common rule of about four characters per token in English, then adjusts for what the text is made of: punctuation, as in code and JSON, splits into more tokens than letters do, and each Chinese, Japanese or Korean character weighs about one token. Use it to size a prompt or a document. For billing, count with your model provider’s own tokenizer.
Why count tokens.
Everything an Agent reads sits in its context window, and the window has a size. Knowing that a document takes 3,000 tokens and not 30,000 tells you whether to paste it, summarize it or let the Agent fetch only the part it needs.
Private by design.
Everything runs in your browser. The text you paste is never sent to a server.
Questions.
What is a token?
A token is the unit a language model reads and writes: a word, part of a word or a punctuation mark. A short common word is one token, and a long or rare word is several.
How many tokens is 1,000 words?
In English, about 1,300, from the rule of roughly three quarters of a word per token. Code, JSON and text in other languages take more.
Is the token count exact?
No, it is an estimate. Each model has its own tokenizer, so the same text gives a different count on different models. The estimate is close enough to size a prompt or a document, and your provider’s tokenizer gives the exact figure for billing.
Does it count characters with or without spaces?
Both. The counts with and without spaces sit side by side.
Is my text uploaded?
No. Every count is calculated in your browser, and the text you paste never leaves your device.
Why would an Agent’s context window matter to me?
An Agent can only work with what fits in its context window. A long document that overflows it is cut or forgotten, so measuring first tells you whether to trim, split or summarize.
More tools.
Give your Agents shared context.
Free plan available, no credit card.