What is a token?
Large language models read text in small pieces called tokens. In English, one token is roughly four characters or three-quarters of a word. Code, numbers and non-English scripts usually use more tokens per character.
How accurate is this estimate?
Every model family uses its own tokenizer, so exact counts differ. This tool blends character- and word-based heuristics, with extra weight for code symbols and non-Latin scripts. Expect it to be within about 10–20% for typical English text. For billing-critical numbers, use your provider's official token-counting API.
Estimating cost
Enter your model's current input price per million tokens (check your provider's pricing page) and how many times you will send this prompt. Output tokens are billed separately, usually at a higher rate.