What does this token counter do?
It estimates how many tokens a piece of text would use for an AI model. Tokens are the small chunks of text that models read and bill for, and they are not the same as words or characters.
Token Counter
Estimate how many AI tokens a piece of text is worth. Paste your text, upload a document, or drag a file in. The estimate is calibrated for OpenAI tokenization, runs in your browser, and nothing is uploaded. Up to 500,000 characters at a time.
Input
0 / 500,000 characters
Token estimate
- Estimated tokens
- 0
- Characters
- 0
- Characters per token
- 0
This is an estimate, not an exact count. Depending on the model and the text, the real token count can differ by a few percent.
Common questions
How accurate is the estimate?
It is calibrated against OpenAI's o200k_base tokenizer. For everyday English, Simplified Chinese, and mixed Chinese and English text, the estimate is usually within a few percent. Treat the number as a close estimate, not an exact count.
Can I upload a file?
Yes. You can upload a single plain-text file such as .txt, .md, .html, or .json, up to 5 MB, or drag a file onto the input box. The file is read locally in your browser and is not uploaded. Binary formats such as PDF, Word, or Excel are not supported.
Which models does it apply to?
The estimate is closest to OpenAI models. Other models, including Claude and Gemini, use different tokenizers, so their real counts can differ more.
When can the estimate be off?
Random or high-entropy strings such as Base64, hashes, and IDs can be underestimated by a lot, because they split into more tokens than ordinary words. Traditional and classical Chinese, as well as some scripts such as Arabic, Hindi, Hebrew, or Thai, can also be underestimated.
Is my text uploaded?
No. The estimate is calculated in your browser, and any file you upload is read locally. No account is required, and your text is not sent to or stored on a server.