A token is the unit of text a large language model reads and writes, and it is usually shorter than a word, so a block of text almost always contains more tokens than words. The estimator below counts the characters and words in your text and converts them into an approximate token count using the two heuristics that hold for typical English: about one token for every four characters, and about three tokens for every four words.
What a Token Is in an LLM
A token is a piece of text that a language model treats as a single unit. A tokenizer splits text into these pieces before the model reads it, and it joins them back into characters when the model writes. A token can be a whole short word, part of a longer word, a single punctuation mark, or a space attached to the word that follows it. Because common words map to one token while longer or rarer words split into several, the token count of any passage sits above its word count and below its character count.
How to Use It
- Paste or type the text you want to measure into the box.
- Read the headline token estimate, which updates as you type.
- Check the character count and word count below it to see where the estimate comes from.
- Compare the character estimate and the word estimate when you need a range rather than a single number.
- Use the figure to plan against a model’s context window or to forecast API cost.
How Words Map to Tokens
Two rules of thumb cover most English text. The first counts characters: roughly four characters of English make one token, so dividing the character count by four gives a token estimate. The second counts words: a token holds about three quarters of a word on average, so dividing the word count by 0.75 gives a second estimate. The two methods rarely agree exactly, because punctuation, numbers, and spacing affect each one differently, so this tool reports both and takes the midpoint as the headline figure.
Why Token Counts Matter
Two limits in every language model are measured in tokens, not words. The first is the context window, the maximum amount of text the model can read and generate in one exchange. If your prompt plus the expected reply exceeds that window, the model cannot process it, so knowing the token count tells you whether your input fits. The second is cost. Most LLM APIs bill per token for both the text you send and the text you receive, so an estimate lets you forecast the price of a request before you run it. Counting tokens turns both limits from guesswork into a number you can plan around.
| Words | Approx. tokens | Typical use |
|---|---|---|
| 1 word | ~1.3 tokens | A single term or keyword |
| 10 words | ~13 tokens | A short sentence |
| 100 words | ~133 tokens | A short paragraph |
| 750 words | ~1,000 tokens | About one page of text |
| 3,000 words | ~4,000 tokens | A long document or prompt |
When to Use It
Reach for a token estimate whenever a length limit or a per-token price is involved. Use it to confirm a prompt fits inside a model’s context window before you send it, to size a document you plan to summarize, to budget the cost of a batch of API calls, or to trim text down to a target token count. It is also useful when you write prompts, because a tighter prompt that says the same thing in fewer tokens leaves more room for the model’s reply and costs less to run.
The Approximation Caveat
This tool gives an estimate, not an exact count. Every model uses its own tokenizer, and those tokenizers split text in different ways, so the same passage can produce different token totals across models. The four-characters-per-token and 0.75-words-per-token figures describe average English prose; text in other languages, programming code, long numbers, or dense punctuation can land well above or below them. When a request sits right at the edge of a context window or a budget, treat the estimate as a guide and confirm the exact count with the tokenizer of the model you are using.
Last Thoughts on Estimating Tokens
Tokens are the real currency of a language model: they set how much text fits and how much each request costs. Words are easy to count by eye, but the model never sees words, so converting words to tokens with a simple multiplier is the fastest way to plan against the limits that actually apply.
Estimate the tokens in your next prompt before you send it, and pair this with our word counter to get the underlying word count, the LLM cost calculator to turn tokens into a price, and the rest of our free online tools.
Key Takeaways:
- A token is the unit an LLM reads and writes; it is usually shorter than a word, so text holds more tokens than words.
- Two heuristics cover English: about 4 characters per token, and about 0.75 words per token (words times 1.33).
- A 100 word passage is roughly 130 to 140 tokens, and about 750 words is close to 1,000 tokens.
- Token counts decide whether text fits a model’s context window and how much an API request costs.
- This estimator runs entirely in your browser; nothing you paste is sent anywhere.
- Exact counts depend on the model’s tokenizer, so use the estimate as a guide, not a final figure.
Frequently Asked Questions (FAQs)
How many tokens is one word?
On average, one English word is about 1.3 tokens, because a token holds roughly three quarters of a word. Short common words map to a single token, while longer or rarer words split into two or more. To convert words to tokens, multiply the word count by about 1.33.
How do I convert words to tokens?
Divide the word count by 0.75, or multiply it by about 1.33. For example, 300 words is roughly 400 tokens. You can also estimate from characters by dividing the character count by four. This tool does both and reports the midpoint as the headline figure.
Is this token count exact?
No. It is an estimate based on average English text. Each model uses its own tokenizer, so the same passage can produce a different total across models. Code, other languages, numbers, and heavy punctuation move the figure further from the average, so confirm the exact count with the model’s tokenizer when it matters.
Why do tokens matter for LLMs?
Tokens set two limits. The context window caps how many tokens a model can read and generate at once, and most APIs bill per token for both input and output. Knowing the token count tells you whether your text fits and what a request will cost before you run it.
How many tokens is 1,000 words?
About 1,300 to 1,400 tokens for typical English prose. The reverse also holds: 1,000 tokens is close to 750 words, or roughly one page of text. Use these as planning figures rather than exact counts.
Does this tool send my text anywhere?
No. The text you paste is measured entirely inside your browser, and it is never uploaded, logged, or stored on any server. You can disconnect from the internet and the estimator will still work.
