Claude token counter
Paste a prompt to estimate its size and cost on Anthropic's Claude models. Claude's tokenizer is not published, so counts are estimates; for exact numbers use Anthropic's token-counting API.
How to use the Claude token counter
- Paste your prompt or document.
- Read the token estimate in the Claude rows of the table.
- Set the expected answer length and the number of requests.
- Compare Opus, Sonnet and Haiku side by side.
What you get
- Token estimates for current Claude models, marked as estimates
- Cost per request and for many requests
- Context window check for each model
- Prices you can edit to match your plan
About the Claude token counter
Why an estimate
Anthropic does not publish Claude's tokenizer for offline use. The estimate uses the same text counted with OpenAI's tokenizer, which is usually within about 20% for English. Code and other languages can differ more.
Good enough for planning
For budgeting, comparing models and checking that a document fits, an estimate is enough. For billing-level accuracy, call Anthropic's count_tokens endpoint with your real request.
Frequently asked questions
- Does Claude use more tokens than GPT?
- Often a little more for the same English text, because the tokenizers differ. The difference depends on the language and content.
- Are the prices current?
- Yes. Prices load live and update every few hours; the time of the last update is shown under the table. Edit a price if your plan differs.
- Is my prompt uploaded?
- No. Everything runs in your browser.