AI Text to Tokens Estimator (characters or words → tokens & cost, by language)

How it works

Tokens ≈ characters ÷ characters per token, or words ÷ words per token, then × (1 + tokenizer adjustment). For English, OpenAI and Google both give about 4 characters per token, and OpenAI about ¾ of a word per token. No provider publishes an official ratio for Korean, Japanese or Chinese, so we use a cautious planning assumption of 1 character per token; real counts vary by model and text. This is an estimate — for exact numbers use the provider’s token-counting tool or the usage figures in your API responses.

⚠ Estimates for information only — not financial advice.

FAQ

Why do Korean and Japanese use more tokens?

Tokenizers are trained mostly on English text, so common English words are single tokens while Korean syllables or Japanese characters are often split into one or more tokens each. OpenAI notes that non-English text usually has a higher token-to-character ratio.

How do I get the exact count?

Send a sample through the provider’s token-counting endpoint or look at the usage numbers returned with each API response, then enter your own characters-per-token ratio here.

Checked 10 Nov 2026. OpenAI Help Center, “What are tokens and how to count them?” (1 token ≈ 4 characters ≈ ¾ word in English); Google AI for Developers, “Understand and count tokens” (about 4 characters per token for Gemini); Anthropic pricing page (Claude 4.7+ tokenizer ≈ 30% more tokens). The Korean/Japanese/Chinese ratio is our own cautious assumption.