Tool
AI Token Counter & Cost Estimator
Paste your prompt, your document, your code — see its token count, how much of the context window it eats, and what the API call actually costs on GPT, Claude or Gemini. Everything runs in your browser; your text never leaves the page.
Estimate
Estimate method: 1.33 tokens per English word, 0.6 per CJK character, 0.25 per other character. Real tokenizers (BPE) vary by model and language; treat this as a budgeting estimate, not a billing figure. Prices are indicative list prices per 1M tokens, checked 2026-11-04 — providers reprice often, so verify before budgeting a product on them.
How the estimate works
Models don't read words or characters — they read tokens, chunks of text learned from training data. English averages about 0.75 words per token; Chinese is denser per character (about 0.6 tokens per character in our model); code and punctuation fall in between. That is why “1000 words” is not a fixed token number: the same word count in Chinese, English, or code tokenizes differently, and the table above shows the breakdown so you can see which part of your text is expensive.
Why cost follows tokens, not words
Every API bill is tokens × price, split into input and output — and output tokens cost 5× input tokens on the current models. A prompt that is 10,000 tokens in and 2,000 out on GPT-6.1 Sol costs about $0.04 at list prices; the same call on Claude Sonnet 5.5 costs about $0.04. When a prompt feels “expensive”, it is almost always the output length or the model choice, not the input length — the calculator makes both visible before you send anything.
Next: the context window table for every major model, or 8 ways to cut token usage without hurting quality.