LLM Token Counter & Word Counter
Paste text to see its character, word and line counts along with an approximate token count for large language models. Switch between GPT and Claude estimate modes to size a prompt before you send it. The token figure is a heuristic based on word and character counts, not an official tokenizer, so real counts vary by model; use each provider's tokenizer for billing. Counting runs in your browser, and your text is never uploaded.
⚠️ Token counts are heuristic estimates, not an official tokenizer. For billing and context limits, rely on the official tokenizers from OpenAI, Anthropic, and others.
How to use the LLM Token Counter
- Paste or type text into the Input text box.
- Choose an estimate mode: GPT estimate or Claude estimate.
- Read the approximate token count along with characters, words and lines. Click Clear to start again.
Frequently asked questions
- It is an estimate: English is counted at roughly 1.25–1.3 tokens per word, and Chinese, Japanese or Korean at roughly 1.5–1.7 tokens per character. That is good enough to gauge prompt size, but real counts depend on each model's tokenizer.
- Both are estimates that use slightly different multipliers to reflect how each tokenizer tends to behave. Neither is an official count.
- Each space-separated chunk counts as one word, and each Chinese, Japanese or Korean character counts as one word.
- No. Counting happens in your browser, and the text is gone when you close the tab.
Other tools
Looking for more AI and software tools? Browse the yunodb.com tools directory ↗.