# Token counter

Count tokens for AI models, with a real tokenizer

Use it at https://www.quicktoolz.com/tools/token-counter. From QuickToolz.

AI models read text as tokens, which are often parts of words. English averages about 4 characters a token. This uses OpenAI’s real o200k_base tokenizer, so that count is exact. Most other companies don’t publish theirs, so their counts are marked as estimates. Your text stays in your browser and is never uploaded.

## How it works

Text is split into tokens, which are often parts of words.

The tokenizer loads when you start typing. The share of a context window shows how much of a model’s limit your text would use.

Worked example: “hello world” is 2 tokens with the o200k_base tokenizer. English text averages about 4 characters a token, and other languages often use more tokens.

## Examples

| Tokenizer | How we count |
| --- | --- |
| o200k_base (OpenAI) | Exact |
| Other models | ≈ estimate, about 4 characters a token |

## What this tool doesn't cover

- Counts for models other than OpenAI’s are estimates.
- Real requests add a few tokens for formatting.

## Questions

### Why do Hindi and other languages use more tokens?

Tokenizers learn from mostly English text, so other scripts split into more pieces.

### What is a context window?

The most text a model can consider at once, measured in tokens.

### Is this exact for every model?

No. It’s exact for o200k_base, and an estimate for others. We label which is which.

### What is a token?

A piece of text that AI models read, often part of a word. In English, one token is about four characters.

### How many tokens is 1,000 words?

About 1,300 in English, though it depends on the words and the model.

## Sources

- [OpenAI tiktoken](https://github.com/openai/tiktoken)