Tokens to Words Calculator
Tokens are pieces of words, not words. This converts in both directions and tells you whether what you have fits in a context window.
How the math works
English prose runs about 1.33 tokens per word, or roughly 4 characters per token. Code averages closer to 1.6 and non-English text often over 2. Words here are 5.5 characters including the space, and a page is 500 words. These are averages; a real tokenizer is exact and this is not.
Common questions
- How many words is 1,000 tokens?
- About 750 words of English prose. The old rule of thumb is that a token is roughly four characters, or three quarters of a word, which holds well for ordinary writing.
- Why does non-English text use more tokens?
- Because the tokenizers were trained mostly on English, so English words often map to a single token while other languages split into several pieces. The same sentence in Japanese or Arabic can cost two or three times as much.
- Does the whole context window belong to my input?
- No. Input, the conversation history and the generated output all share it. Leave headroom, and remember that in a long chat the history grows every turn until it crowds out everything else.