Context windows and prices are quoted in tokens. Knowing roughly how many tokens a prompt uses helps you stay within limits and budget.
What is a token?
A token is a chunk of text that a model reads as one unit: a short word, part of a longer word, or a piece of punctuation. In English, a token is on average about four characters or three quarters of a word. See the glossary entry on tokens.
How the estimate works
The Token Counter gives two estimates and the raw counts. For “The quick brown fox jumps over the lazy dog.” (44 characters, 9 words), it estimates 11 tokens by characters (44 ÷ 4) and 12 by words (9 ÷ 0.75). Real tokenizers will usually be in that range for ordinary English, but not exactly.
Using the tool
- Paste your text.
- Read both estimates, the character count and the word count.
- Use the higher figure when you are close to a limit.
When the estimate is off
- Code, numbers and unusual words use more tokens per character.
- Languages other than English often need more tokens for the same meaning.
- Different models use different tokenizers, so the same text can count differently.
The tool does not run a model’s real tokenizer. For exact counts, use the one provided by your model’s maker. To turn a token count into a cost, see how to estimate LLM API cost.
Frequently asked questions
Is the text sent anywhere?
No. The estimate runs in your browser.