1 million tokens

1 Million Tokens to Words

What does a million tokens actually hold? Convert any token count to words and see what fits inside.

Short answer

1 million tokens ≈ 750,000 English words — about 7–8 full-length novels.

Based on ~0.75 words per token for English prose. Code, tables, and non-English text hold fewer words per million tokens.

Tokens

Estimated words

≈ 750,000

1.00M tokens × 0.75 words/token

Token counts at human scale

TokensRoughly equals
1,000 tokens750 words — a couple of pages
4,000 tokens3,000 words — a long article
8,000 tokens6,000 words — a short story
32,000 tokens24,000 words — a novella
128,000 tokens96,000 words — a short novel
200,000 tokens150,000 words — 2–3 novels
1,000,000 tokens750,000 words — 7–8 novels

Comparisons assume a ~100,000-word novel and ~500 words per page. Newer tokenizers and non-English text shift these numbers.

FAQ

How many words is 1 million tokens?

Roughly 750,000 English words. That is the widely used rule of thumb (0.75 words per token) for ordinary prose.

How many books is 1 million tokens?

About 7–8 full-length novels, assuming ~100,000 words per novel. It is also roughly 1,500 printed pages at 500 words per page.

Does 1 million tokens mean the same for every model?

The word count is roughly comparable, but each model family uses its own tokenizer, so the exact word equivalent shifts a little. Newer tokenizers are generally more efficient, especially outside English.

Which models have a 1 million token context window?

Claude Fable 5.1, Opus 5, and Sonnet 5, several GPT-5.x models, and Gemini models all offer around 1M input tokens. Check the context window checker to compare exact windows.

How much does 1 million tokens cost?

It depends on the model. At $2.50 per million input tokens, 1M tokens costs $2.50; at $0.30 it costs $0.30. Use the token cost calculator for exact per-model pricing.

What can you fit in 1 million tokens?

A large monorepo's source code, several novels, a full legal contract suite, or months of conversation history — in a single prompt.