Token Counter
Paste text to count its tokens and check whether it fits a model's context window.
Cheapest to Send Once
| # | Model | Cost to send this text once (input only) | Context |
|---|
Counts use the selected tokenizer for every model, so non-OpenAI models are approximate. OpenAI tokenizers give exact counts. Output tokens are not included.
What It Is For
See how many tokens a prompt or document uses and whether it fits a model's context window, before you send it. Planning a whole feature rather than measuring one text? Use the Cost Forecaster to turn it into a monthly budget. The forecaster opens with your token count and typical defaults for output size and volume, which you can change.
Questions
Is my text uploaded?
No. Counting happens in your browser.
How accurate is the count?
Exact for OpenAI models when the tokenizer loads. Anthropic and Google do not publish tokenizers that run offline, so those counts are estimates, usually within 10 to 15 percent.
Where do the prices come from?
From the open-source LiteLLM catalog, refreshed daily. They are not verified against provider billing. The last-verified date is shown above the table.