Tokens to Words Converter
How many words is 1,000 tokens? About 750, in plain English prose. Use the two-way converter below to translate any token count into words and pages — and see why prompts, transcripts and code all convert differently.
Tokens to words converter
1,000 tokens ≈ 750 words (≈ 1.5 pages at 500 words per page)
Rule of thumb for English prose: 1 token ≈ 0.75 words ≈ 4 characters. Code, non-English languages and unusual formatting can shift the ratio by 30% or more — see the examples below.
Quick conversion table
| Tokens | ≈ Words | ≈ Pages (500 words) | What that looks like |
|---|---|---|---|
| 1,000 | 750 | 1.5 | A long email or a short blog post |
| 10,000 | 7,500 | 15 | A detailed essay or a product spec |
| 100,000 | 75,000 | 150 | A short book — or one full long-context window |
| 1,000,000 | 750,000 | 1,500 | Roughly ten novels; a month of traffic for a small chatbot |
How the conversion works
words ≈ tokens × 0.75tokens ≈ words ÷ 0.75 (≈ words × 1.33)
LLMs don't read words — they read tokens, subword pieces produced by the model's tokenizer. Everyday English words like cost or model are one token each, while longer or rarer words are split into several pieces. Averaged over normal prose that comes out near 4 characters, or 0.75 words, per token — the factor this converter uses. It is a planning estimate, not an exact count: every model family tokenizes slightly differently, and the only exact number comes from running the actual tokenizer.
Worked example. A 1,000-token API response converts to about 1,000 × 0.75 = 750 words — a page and a half at 500 words per page. In reverse, a 1,500-word article is about 1,500 ÷ 0.75 = 2,000 tokens. If that same 2,000 tokens is Python code instead of prose, expect it to hold only 600–1,000 words' worth of source, because symbols and whitespace eat tokens.
What token counts look like in practice
Prompts. A casual chat question is 20–100 tokens (15–75 words). A production system prompt with instructions, rules and examples typically runs 500–2,000 tokens — that's 375–1,500 words you re-send with every single request, which is why prompt length shows up so directly in monthly bills.
Documents. A ten-page contract at 500 words per page is about 5,000 words ≈ 6,700 tokens. A typical RAG chunk of 500 tokens carries roughly 375 words — a couple of solid paragraphs.
Transcripts. People speak at roughly 130 words per minute, so one hour of meeting audio is about 7,800 words ≈ 10,400 tokens. A full 8-hour workshop day lands near 83,000 tokens — close to the limit of a 100K-context model before you even add instructions.
Code. Source code converts worst: identifiers, brackets and indentation push it to 0.3–0.5 words per token. A 300-line Python file often costs 3,000–4,000 tokens, and minified JSON is denser still.
Once you know your token count, the numbers turn into money: the LLM Token Cost Calculator converts monthly token volume into an API bill across every major model, and the guide on estimating monthly LLM token usage shows how to get from requests per day to that volume in the first place.
Frequently asked questions
Related tools and guides
- LLM Token Cost Calculator — turn your token count into a monthly API bill, for every major model
- How to estimate your monthly LLM token usage — from requests per day to tokens per month
- AI API Budget Calculator — budget an AI feature from users and requests per user
- LLM Pricing Comparison — what those tokens cost per provider, side by side