LLM Token Estimator
Paste text and see roughly how many tokens it is, as a range. Every model counts tokens differently, so this is an estimate. Your text stays in your browser and is never sent anywhere.
The text is not put in the URL either. Copying the link only carries the window size.
Your estimate will appear here
Your estimate will appear here
English uses Google's official rule of thumb: a token is about 4 characters, and 100 tokens are about 60 to 80 English words. There is no official ratio for Korean or Han characters, so a syllable is taken as roughly 0.5 to 1.4 tokens depending on the tokenizer.
Frequently asked questions
Why is the result a range instead of one number?
Each model uses its own tokenizer, so the same text has a different count on different models. This tool does not run any model's tokenizer. It estimates from ratios per character type, so it shows a low and a high estimate.
Does Korean use more tokens than English?
Usually, for the same meaning. An English word often fits in one or two tokens, while a Korean syllable is often around one token (a rough tendency, not an official figure). Tokenizers keep changing, so compare with the official tools for an exact answer.
Where can I get the exact count?
OpenAI points to its tiktoken library, and Anthropic and Google provide token counting APIs. The official links below show how to count for your model.
Is my text stored anywhere?
No. The estimate runs in your browser, and the text is never sent to a server or put in the URL.
Researchlog collects official AI announcements as summary cards
Only the operator uses it for now, so the updates screen stays closed even after sign-up. You can see what it does first.