How Many Words Is 1 Million Tokens?
By TokenMath Editorial & Engineering Team · Reviewed · Methodology
Convert Tokens to Words
Convert between tokens and words for prose, technical writing or code (also available as a standalone tokens to words converter). For an exact count for a specific model, paste your text into the AI token counter.
Tokens to Words Conversion Table
Based on 1.33 tokens per word for standard English prose and 500 words per single-spaced page:
| Tokens | ≈ English Words | ≈ Pages | ≈ Characters |
|---|---|---|---|
| 100 | 75 | < 1 | 400 |
| 500 | 376 | < 1 | 2,000 |
| 1,000 | 752 | 2 | 4,000 |
| 4,000 | 3,008 | 6 | 16,000 |
| 8,000 | 6,015 | 12 | 32,000 |
| 10,000 | 7,519 | 15 | 40,000 |
| 32,000 | 24,060 | 48 | 128,000 |
| 100,000 | 75,188 | 150 | 400,000 |
| 128,000 | 96,241 | 192 | 512,000 |
| 200,000 | 150,376 | 301 | 800,000 |
| 500,000 | 375,940 | 752 | 2,000,000 |
| 1,000,000 | 751,880 | 1,504 | 4,000,000 |
| 2,000,000 | 1,503,759 | 3,008 | 8,000,000 |
How Many Words Is 1 Million Tokens?
One million tokens is roughly 750,000 words of English prose. To put that in perspective, it's about 1,500 single-spaced pages, the full text of 8 to 10 average novels, or the complete source of a mid-sized codebase. Most current frontier models (Claude Sonnet 5, GPT-5.6, Gemini 3.8) accept around 1 million tokens of context in a single request — see the context window guide for what that means for cost and quality.
Providers price per 1 million tokens, so this is also the unit you'll see on every pricing page. Use the LLM cost calculator to turn a token count into a monthly bill.
1. The Baseline English Prose Ratio
For standard grammatical English text (articles, essays, emails):
2. Source Code, JSON & Technical Documents
Code does not follow natural language distribution. Identifiers like fetchCustomerSubscriptionHistory get fragmented into 4 or 5 subwords. In addition, syntax characters like braces ({ }), tabs, and semicolons often consume independent tokens.
- Python / TypeScript: 1 line of code typically averages 10 to 14 tokens.
- JSON Payloads: Repeating keys (
"status","created_at") and quotes cause high token inflation. A 10KB JSON payload is usually 2,800 to 3,500 tokens. - Single-Spaced PDF Page: A standard 8.5" x 11" single-spaced page (approx. 500 words) produces approximately 650 to 750 tokens.
3. Multilingual Multipliers
Because tokenizers dedicate a majority of vocabulary entries to English, non-Latin alphabets suffer from byte fallback penalties:
| Language Script | Tokens Per 1,000 Words | Cost Multiplier vs English |
|---|---|---|
| English | ~1,330 tokens | 1.0x (Baseline) |
| Spanish / French / German | ~1,500 - 1,700 tokens | 1.1x - 1.3x |
| Russian (Cyrillic) | ~2,200 - 2,600 tokens | 1.7x - 2.0x |
| Japanese / Chinese / Korean | ~2,400 - 3,200 tokens | 1.8x - 2.4x |
4. How Are AI Tokens Calculated?
Before a model sees your text, a tokenizer splits it into pieces from a fixed vocabulary of 65,000 to 256,000 entries. Algorithms like Byte-Pair Encoding (BPE) start from single bytes and repeatedly merge the most frequent pairs, so common words ("the", "engineering") become one token while rare words split into several. That's why the tokens-per-word ratio changes with the content and the model. The tokenization guide explains BPE, WordPiece and SentencePiece in detail.
Billing then counts two numbers: input tokens (everything you send, including system prompts and chat history) and output tokens (everything the model generates, including hidden reasoning). See input vs output tokens for why output costs 3–5x more.
5. Why Different Models Give Different Counts
The same paragraph can be 5–10% more tokens on one model than another. OpenAI's o200k_base has a ~200k vocabulary; Claude's is ~65k; Gemini's SentencePiece vocabulary is ~256k. Larger vocabularies encode longer strings per token. Check a specific model with the GPT, Claude or Gemini token counters.
Tokens vs Words FAQ
How many words is 1 million tokens?
About 750,000 English words, or roughly 1,500 single-spaced pages. That's around 8 to 10 average-length novels. For source code or JSON, 1 million tokens holds closer to 450,000 to 600,000 words because symbols and identifiers use extra tokens.
How many words is 100k tokens?
About 75,000 English words, or roughly 150 single-spaced pages — the length of a typical novel.
How many tokens per word?
English prose averages about 1.33 tokens per word (1 token ≈ 0.75 words ≈ 4 characters). Technical writing averages about 1.6 tokens per word and source code about 2.2.
How are AI tokens calculated?
A tokenizer splits text into subword pieces from a fixed vocabulary using algorithms like Byte-Pair Encoding (BPE) or SentencePiece. Common words become one token; rare words, numbers and code identifiers are split into several. The provider then bills you per 1 million input and output tokens.
How do I convert words to tokens?
Multiply the word count by 1.33 for English prose (e.g. 1,000 words ≈ 1,330 tokens). For an exact count for a specific model, paste the text into a token counter.
Use the official TokenMath CLI with zero installation: cat prompt.txt | npx @tokenmath/tokenmath count or read our Terminal CLI documentation.