How Many Tokens Is It?
1,000 words is about 1,333 tokens. One token is roughly 0.75 English words, so multiply a word count by 1.33 to get tokens. A single-spaced page (about 500 words) is about 667 tokens, and a line of code averages about 10.
Want an exact count for your own text? Paste it into the token counter. To go the other way, from tokens to words, use the tokens to words converter.
Words to tokens
Tokens ≈ words × 1.33. Costs are for sending the text as input, at prices checked 2026-10-02.
| Words | Tokens | GPT-6 Luna input | Claude Sonnet 5 input | Claude Opus 5.5 input |
|---|---|---|---|---|
| 100 words | ≈ 133 | <$0.0001 | $0.0003 | $0.0005 |
| 250 words | ≈ 333 | <$0.0001 | $0.0007 | $0.0013 |
| 500 words | ≈ 667 | <$0.0001 | $0.0013 | $0.0027 |
| 1,000 words | ≈ 1,333 | $0.0001 | $0.0027 | $0.0053 |
| 2,000 words | ≈ 2,667 | $0.0003 | $0.0053 | $0.01 |
| 3,000 words | ≈ 4,000 | $0.0004 | $0.0080 | $0.02 |
| 5,000 words | ≈ 6,667 | $0.0007 | $0.01 | $0.03 |
| 10,000 words | ≈ 13,333 | $0.0013 | $0.03 | $0.05 |
| 20,000 words | ≈ 26,667 | $0.0027 | $0.05 | $0.11 |
| 50,000 words | ≈ 66,667 | $0.0067 | $0.13 | $0.27 |
| 75,000 words | ≈ 100,000 | $0.01 | $0.20 | $0.40 |
| 100,000 words | ≈ 133,333 | $0.01 | $0.27 | $0.53 |
| 250,000 words | ≈ 333,333 | $0.03 | $0.67 | $1.33 |
| 500,000 words | ≈ 666,667 | $0.07 | $1.33 | $2.67 |
| 1,000,000 words | ≈ 1,333,333 | $0.13 | $2.67 | $5.33 |
Pages to tokens
A single-spaced page of about 500 words is roughly 667 tokens. These are pages of extracted text; a PDF read with its layout and images costs more — seehow many PDF pages fit in 1 million tokens.
| Pages | Tokens | GPT-6 Luna input | Claude Sonnet 5 input | Claude Opus 5.5 input |
|---|---|---|---|---|
| 1 page | ≈ 667 | <$0.0001 | $0.0013 | $0.0027 |
| 5 pages | ≈ 3,335 | $0.0003 | $0.0067 | $0.01 |
| 10 pages | ≈ 6,670 | $0.0007 | $0.01 | $0.03 |
| 20 pages | ≈ 13,340 | $0.0013 | $0.03 | $0.05 |
| 50 pages | ≈ 33,350 | $0.0033 | $0.07 | $0.13 |
| 100 pages | ≈ 66,700 | $0.0067 | $0.13 | $0.27 |
| 500 pages | ≈ 333,500 | $0.03 | $0.67 | $1.33 |
| 1,000 pages | ≈ 667,000 | $0.07 | $1.33 | $2.67 |
Lines of code to tokens
Code averages about 10 tokens per line. Brackets, indentation and long identifiers push it up; terse languages come in lower.
| Lines of code | Tokens | GPT-6 Luna input | Claude Sonnet 5 input | Claude Opus 5.5 input |
|---|---|---|---|---|
| 100 lines of code | ≈ 1,000 | $0.0001 | $0.0020 | $0.0040 |
| 500 lines of code | ≈ 5,000 | $0.0005 | $0.01 | $0.02 |
| 1,000 lines of code | ≈ 10,000 | $0.0010 | $0.02 | $0.04 |
| 5,000 lines of code | ≈ 50,000 | $0.0050 | $0.10 | $0.20 |
| 10,000 lines of code | ≈ 100,000 | $0.01 | $0.20 | $0.40 |
| 50,000 lines of code | ≈ 500,000 | $0.05 | $1.00 | $2.00 |
| 100,000 lines of code | ≈ 1,000,000 | $0.10 | $2.00 | $4.00 |
How the conversion works
AI models don't read words; they read tokens — whole words, pieces of words, or punctuation. In typical English, one token covers about 0.75 words or 4 characters, which is where the ×1.33 multiplier comes from.
The ratio shifts with the text and the model. Rare words, numbers, code and most non-English languages split into more tokens. Each provider also uses its own tokenizer: Anthropic notes that its tokenizer for Claude 4.7 and later models produces about 30% more tokens for the same text than its earlier one. Treat these tables as planning estimates and count real text when the budget matters.
Frequently asked questions
- How many tokens is 1,000 words?
- About 1,333 tokens, using the standard ratio of 1 token per 0.75 English words.
- How many tokens is one page?
- About 667 tokens for a single-spaced page of roughly 500 words. A double-spaced page is about half that.
- How many tokens is 100 lines of code?
- About 1,000 tokens. Code averages roughly 10 tokens per line, depending on the language and how verbose its identifiers are.
- Are these counts exact?
- No — they are estimates from average ratios. Each model family has its own tokenizer, so the same text produces different counts. Paste your text into the token counter for an exact OpenAI count and an estimate for Claude and Gemini.