Calculatte

HomeGuides › How Many Tokens Is My Text? A Quick Estimation Guide

How Many Tokens Is My Text? A Quick Estimation Guide

How to estimate AI tokens in a piece of text using the standard character and word rules of thumb, with conversion tables and common context window sizes.

Updated · 2 min read

In this guide
  1. The rule of thumb
  2. Words to estimated tokens
  3. Why it is only an estimate
  4. Context windows in plain terms
  5. A worked example
  6. Try it yourself

Quick answer: for English text, one token is roughly 4 characters, or about 0.75 words. A 500-word page is therefore roughly 667 tokens. Paste your own text into the token calculator for an instant estimate.

The rule of thumb

AI language models such as GPT and Claude do not read whole words. They break text into smaller pieces called tokens, which can be a whole word, part of a word, or a single character or symbol. OpenAI publishes a simple approximation for English text: about 4 characters per token, or about 100 tokens per 75 words. Neither figure is exact, but both land close enough for planning purposes.

Words to estimated tokens

WordsEstimated tokens
100≈ 133
250≈ 333
500≈ 667
1,000≈ 1,333
1,500≈ 2,000
2,000≈ 2,667
5,000≈ 6,667
10,000≈ 13,333

To estimate any word count yourself, divide by 0.75, or equivalently, multiply by roughly 1.33.

Why it is only an estimate

  • Common English words are often a single token, while rare words, names and made-up words can split into two or more.
  • Numbers, code and heavy punctuation typically use more tokens per character than plain prose.
  • Languages other than English, especially those not written in the Latin alphabet, often need noticeably more tokens for the same idea.
  • Different providers use different tokenizers, so the same text can come out a little different on GPT versus Claude versus another model.

Context windows in plain terms

A model's context window is the total number of tokens it can hold in a single conversation, combining what you send and what it generates back. Once a conversation grows past that limit, the oldest content has to be dropped or summarized. Here is roughly how far common context window sizes stretch:

Context window≈ Words≈ Pages (500 words/page)
4,000 tokens (a short window)3,0006
8,000 tokens (a common short window)6,00012
32,000 tokens (a mid-sized window)24,00048
128,000 tokens (a long window)96,000192
200,000 tokens (a very long window)150,000300
1,000,000 tokens (an extra-large window)750,0001,500

A worked example

A 1,500-word blog post is roughly 1,500 ÷ 0.75 = 2,000 tokens. That is a small fraction of even a short 8,000-token context window, so length alone is rarely the limiting factor for a single article, it becomes relevant when a conversation, a long document, or a big pasted codebase adds up over many turns.

Try it yourself

The AI token calculator applies this same rule of thumb to any text you paste in, and shows what share of common context windows it would use. The word counter also shows a quick token estimate alongside its other stats.

Try the calculators

General information only, not professional advice. See our editorial policy.