LLM Token Counter and Cost Calculator
Paste a prompt, see how many tokens it costs and what that price becomes across the models you are actually choosing between.
Why token counts matter
Model pricing is per token, and a token is roughly four English characters — but not always. Code, tables and non-Latin scripts pack more meaning into fewer characters, so the same visible text can cost noticeably more.
This tool uses a character-based estimate. It is accurate enough for budgeting and ranking models against each other; it is not a billing-grade count. Providers expose an exact tokenizer endpoint if you need the real number.
Frequently asked questions
How many characters is one token?
About four characters of English text on average. Chinese, Japanese and Korean characters are typically one token each, which is why a character counter badly underestimates their cost.
Are output tokens charged separately?
Yes, and they cost more than input on most models. A 10-turn conversation is charged 10 times on the input plus every response.
Does the context window affect cost?
It affects what fits, not the price. You pay for what you send, whether that is 200 or 200,000 tokens.
Are the prices accurate?
They are the published list prices at the time of writing and are stored in one editable block in the tool. Always confirm against your provider dashboard before committing.
Is my prompt sent anywhere?
No. Counting happens in this page and your text is never uploaded.