Home › Tools › Pages to tokens
Pages to tokens converter
A standard page is 500 words, which is about 665 tokens. Use this when you are estimating the cost of feeding documents, manuals or contracts to a model.
Pages to tokens table
| Pages | Tokens |
|---|---|
| 1 | 665 |
| 2 | 1,330 |
| 5 | 3,325 |
| 10 | 6,650 |
| 20 | 13,300 |
| 50 | 33,250 |
| 100 | 66,500 |
| 250 | 166,250 |
| 500 | 332,500 |
| 1,000 | 665,000 |
What 10 pages costs on every model
| # | Model | Provider | Input | Output | Blended |
|---|---|---|---|---|---|
| 1 | GPT-6 Luna | OpenAI | $0.00067 | $0.00332 | $0.0015 |
| 2 | GPT-5.6 Luna | OpenAI | $0.00133 | $0.00798 | $0.00332 |
| 3 | GPT-5.4 nano | OpenAI | $0.00133 | $0.00831 | $0.00341 |
| 4 | Gemini 3.5 Flash-Lite | Google Gemini | $0.00199 | $0.0166 | $0.00615 |
| 5 | DeepSeek V4.1 Flash | DeepSeek | $0.00199 | $0.00798 | $0.00399 |
| 6 | GPT-5.4 mini | OpenAI | $0.00499 | $0.0299 | $0.0125 |
| 7 | Gemini 3.8 Flash | Google Gemini | $0.00499 | $0.0249 | $0.0112 |
| 8 | Gemini 3.7 Flash | Google Gemini | $0.00499 | $0.0249 | $0.0112 |
| 9 | Claude Haiku 4.5 | Anthropic | $0.00665 | $0.0333 | $0.015 |
| 10 | Gemini 2.5 Pro | Google Gemini | $0.00831 | $0.0665 | $0.0249 |
| 11 | Grok 4.3 | xAI Grok | $0.00831 | $0.0166 | $0.0125 |
| 12 | DeepSeek V4 Pro | DeepSeek | $0.00878 | $0.0263 | $0.0154 |
| 13 | Gemini 3.5 Flash | Google Gemini | $0.00997 | $0.0599 | $0.0249 |
| 14 | GPT-5.2 | OpenAI | $0.0116 | $0.0931 | $0.0349 |
| 15 | GPT-6.1 Sol | OpenAI | $0.0133 | $0.0665 | $0.0299 |
| 16 | GPT-6 Sol | OpenAI | $0.0133 | $0.0665 | $0.0299 |
| 17 | GPT-5.6 Terra | OpenAI | $0.0133 | $0.0798 | $0.0333 |
| 18 | GPT-4.1 | OpenAI | $0.0133 | $0.0532 | $0.0266 |
| 19 | Claude Sonnet 5 | Anthropic | $0.0133 | $0.0665 | $0.0299 |
| 20 | Gemini 3.1 Pro (Preview) | Google Gemini | $0.0133 | $0.0798 | $0.0333 |
| 21 | Grok 4.6 | xAI Grok | $0.0133 | $0.0399 | $0.0233 |
| 22 | GPT-5.4 | OpenAI | $0.0166 | $0.0997 | $0.0416 |
| 23 | Claude Sonnet 4.6 | Anthropic | $0.0199 | $0.0997 | $0.0449 |
| 24 | GPT-5.6 Sol | OpenAI | $0.0266 | $0.133 | $0.0599 |
| 25 | GPT-5.5 | OpenAI | $0.0333 | $0.199 | $0.0831 |
| 26 | Claude Opus 5 | Anthropic | $0.0333 | $0.166 | $0.0748 |
| 27 | GPT-6 Astra | OpenAI | $0.0665 | $0.332 | $0.15 |
| 28 | Claude Fable 5.1 | Anthropic | $0.0665 | $0.332 | $0.15 |
Input = 6,650 prompt tokens. Output = 6,650 generated tokens. Blended = input + 25% of output, the same weighting used across this site.
Document ingestion, priced
The table above prices a single pass. Retrieval-augmented generation re-sends retrieved chunks on every question, so a 100-page corpus queried 1,000 times does not cost 100 pages of input — it costs whatever the retrieved slice is, multiplied by the number of questions. That is why caching the stable part of the prompt matters more than shrinking the corpus.
FAQ
How many tokens is a page?
About 665 tokens at 500 words per page. A dense A4 page of 650 words runs closer to 865.
How many pages can a 1M-token context hold?
Roughly 1,500 pages. In practice you never use the full window: cost scales with every token, and long-context surcharges apply.
What does indexing 100 pages cost?
At 500 words per page, 100 pages is 66,500 tokens — $0.00665 on the cheapest model in the dataset.
Keep going
- Tokens to pages
- Words to tokens
- Compare caching economics between models
- All 28 models ranked by price
- Run your own volumes in the calculator