Free tool · Converter
Tokens ↔ Words Converter
How many tokens is 1,000 words? About 1,333, by OpenAI's published rule of thumb that 1 token is about 0.75 words of English, and 1,250 to 1,667by Google's published range for Gemini. Type any amount in words, tokens or characters and the rest update instantly, with the fit against each model's context window. For the exact token count of a specific piece of text, use the Token Counter.
- Words to tokens
- words ÷ 0.75
- about × 1.33
- Tokens to words
- tokens × 0.75
- 1 token ≈ 4 characters
OpenAI's published rule of thumb for English text, the one the converter below uses for English.
English prose. 1,333 tokens, 1,000 words, 5,333 characters
Tokens
1,333
Words
1,000
Characters
5,333
How it fits common context windows
- 8K (under every tracked model)17% used
- 32K (under every tracked model)4% used
- 128K (under every tracked model)1% used
- 200K (the smallest tracked window)1% used
- 1M (32 of 42 tracked models)0% used
English uses OpenAI's published rule of thumb: 1 token ≈ 0.75 words ≈ 4 characters. The Code / dense and Other languages settings use our own planning ratios, which no provider publishes. Every result is an estimate: the exact count depends on the specific text and the model's tokenizer.
Need the exact count for specific text?
How many tokens is N words?
Computed from each provider's published rule of thumb for English text, so these are estimates, not tokenizer counts. OpenAI's rule is 1 token ≈ 0.75 words; Google's rule for Gemini is 100 tokens ≈ 60-80 English words, so it gives a range.
| Words (English) | Tokens, OpenAI's rule | Tokens, Google's Gemini rule |
|---|---|---|
| 100 | 133 | 125 to 167 |
| 250 | 333 | 313 to 417 |
| 500 | 667 | 625 to 833 |
| 750 | 1,000 | 938 to 1,250 |
| 1,000 | 1,333 | 1,250 to 1,667 |
| 1,500 | 2,000 | 1,875 to 2,500 |
| 2,000 | 2,667 | 2,500 to 3,333 |
| 3,000 | 4,000 | 3,750 to 5,000 |
| 5,000 | 6,667 | 6,250 to 8,333 |
| 10,000 | 13,333 | 12,500 to 16,667 |
| 50,000 | 66,667 | 62,500 to 83,333 |
| 100,000 | 133,333 | 125,000 to 166,667 |
Anthropic's models overview says 1 million tokens is roughly 555,000 words on its current tokenizer, introduced with Claude Opus 4.7, and about 750,000 words on models before it. Anthropic's glossary says a token on models before Claude 4.7 represents about 3.5 English characters, and that Claude 4.7 and later use a newer tokenizer that produces about 30 percent more tokens for the same text. For an exact Claude count, Anthropic points to its token counting API.
Sources, read 22 October 2026: OpenAI API docs: Key concepts; Gemini API docs: Understand and count tokens; Claude docs: Glossary (Tokens); Claude docs: Models overview.
How many words is N tokens?
The counts people most often arrive with (round numbers) and the context-window sizes printed on model spec sheets. Pages assume 250 words to a paperback page.
| Tokens | Words (English) | Words (code / dense) | ≈ Pages |
|---|---|---|---|
| 1,000 | 750 | 500 | 3 |
| 10,000 | 7,500 | 5,000 | 30 |
| 32,000 (32K) | 24,000 | 16,000 | 96 |
| 100,000 (100K) | 75,000 | 50,000 | 300 |
| 128,000 (128K) | 96,000 | 64,000 | 384 |
| 200,000 (200K) | 150,000 | 100,000 | 600 |
| 1,000,000 (1M) | 750,000 | 500,000 | 3,000 |
English uses OpenAI's published 0.75 words per token; the code / dense column uses our own planning ratio of 0.5, which no provider publishes. Both are the ratios the converter above runs on, so the two cannot disagree. These are rules of thumb, not tokenizer output. For the real count of a specific piece of text, the Token Counter runs the actual GPT tokenizer in your browser, and the Context Window Calculator answers the follow-up question of which models can hold it.
Frequently asked
How do I convert words to tokens?
Divide the word count by 0.75, which is the same as multiplying it by about 1.33. That is OpenAI's published rule of thumb for English text: 1 token is about 4 characters or 0.75 words. To go from tokens to words, multiply by 0.75 instead. Google publishes a range for Gemini, 100 tokens to about 60-80 English words, so its estimate is a range rather than one number. The converter above does the arithmetic in either direction.
How many tokens is 1,000 words?
About 1,333 tokens by OpenAI's published rule of thumb (words ÷ 0.75), and 1,250 to 1,667 tokens by Google's published range for Gemini. Anthropic's models overview says 1 million tokens is roughly 555,000 words on its current tokenizer, introduced with Claude Opus 4.7, and about 750,000 words on models before it. Anthropic's glossary says a token on models before Claude 4.7 represents about 3.5 English characters, and that Claude 4.7 and later use a newer tokenizer that produces about 30 percent more tokens for the same text. For an exact Claude count, Anthropic points to its token counting API. For code, the "Code / dense" setting above uses our own planning ratio, 2,000 tokens for the same thousand words, which no provider publishes. For an exact figure on specific text, use the Token Counter.
How many words is 1 million tokens?
About 750,000 words of ordinary English, roughly 3,000 paperback pages of 250 words each. That figure applies the rule of thumb OpenAI publishes for English text, tokens × 0.75 ≈ words. At our own planning ratio for code, which no provider publishes, the same million tokens carries about 500,000 words.
How many words is 100,000 tokens?
About 75,000 words of English, roughly 300 paperback pages. A 128,000-token window holds about 96,000 words, and a 200,000-token window about 150,000. Both are small by 2026 standards: 128K is under every tracked model, and 200K is the smallest tracked window.
How many words fit in a 128K context window?
About 96,000 words of English (128,000 tokens × 0.75 words per token). A 200K window holds ~150,000 words; a 1M window, which 32 of the 42 models with a published window now reach, holds ~750,000 words, roughly a 3,000-page book. The converter above shows the fit for each common window as you type.
Do ChatGPT, Claude and Gemini count tokens the same way?
No. Each model family has its own tokenizer, and the three providers publish different rules of thumb. OpenAI says 1 token is about 4 characters or 0.75 words of English. Google says a Gemini token is about 4 characters and 100 tokens is about 60-80 English words. Anthropic's models overview says 1 million tokens is roughly 555,000 words on its current tokenizer, introduced with Claude Opus 4.7, and about 750,000 words on models before it. Anthropic's glossary says a token on models before Claude 4.7 represents about 3.5 English characters, and that Claude 4.7 and later use a newer tokenizer that produces about 30 percent more tokens for the same text. For an exact Claude count, Anthropic points to its token counting API.
What's the difference between words, tokens, and characters?
Characters are individual letters/symbols. Words are space-separated. Tokens are the chunks a language model actually reads, usually a short word or a word-piece, and they are what you are billed on and what the context window is measured in. By OpenAI's published rule for English, 1 token is roughly 0.75 of a word, or about 4 characters, but it varies with punctuation, formatting, code, and language.
Is this exact?
No. The English figures come from OpenAI's and Google's published rules of thumb for English text, and the Code / dense and Other languages figures are our own planning ratios, which no provider publishes. The true count depends on the exact text and each model family's tokenizer. It is close enough for planning prompt size, context fit, and rough cost. When you need the precise number for a specific piece of text, paste it into the Token Counter, which runs the real GPT tokenizer in your browser, or use the provider's own token counting tool.