Home › Tools › Characters to tokens
Characters to tokens converter
The rule of thumb is 4 characters per token for English. It is the fastest estimate you can do in your head, and it holds within about 10% for ordinary prose.
4Characters per token
0.25Tokens per character
250Tokens per 1,000 chars
25,000Tokens per 100K chars
Characters to tokens table
| Characters | Tokens |
|---|---|
| 100 | 25 |
| 500 | 125 |
| 1,000 | 250 |
| 2,000 | 500 |
| 5,000 | 1,250 |
| 10,000 | 2,500 |
| 50,000 | 12,500 |
| 100,000 | 25,000 |
| 500,000 | 125,000 |
| 1,000,000 | 250,000 |
Where the 4:1 ratio breaks
| Content type | Characters per token | Why |
|---|---|---|
| English prose | 4.0 | Common words are single tokens |
| Source code | 3.0 | Punctuation and camelCase split identifiers |
| JSON / XML | 2.5 | Repeated quotes and braces tokenise separately |
| Arabic | 1.8 | Diacritics and affixes fragment |
| Chinese | 1.4 | One character is often a full token |
| Base64 / minified code | 1.2 | Random strings have no reusable pieces |
FAQ
How many tokens is 1,000 characters?
About 250 tokens in English. Spaces and punctuation count toward characters but compress well, so the ratio holds across most prose.
Does whitespace count as characters?
Yes. A space is a character, though BPE folds it into the following word, so leading spaces rarely cost an extra token on their own.
What about emoji and Chinese?
Emoji can take two to four tokens each. Chinese averages about 0.7 characters per token, not 4, so divide by 1.5 instead of 4 for CJK text.
Keep going
- Tokens to characters
- Words to tokens
- Arabic token ratios
- All 28 models ranked by price
- Run your own volumes in the calculator