How do I count tokens in a prompt?

Paste the text here. We estimate tokens as the higher of characters÷4 and words×1.3. Official tokenizer counts differ by model.

Is a token the same as a word?

No. English often lands near 0.75 words per token. Code, URLs, and other languages can use many more tokens per word.

Lab · Context windows · ~2 min

Paste text to estimate tokens, see how much of an 8k–1M context window you used, and learn why token counts are not word counts.

Text to count

Paste a prompt, a document, or a chat transcript. Token counts are estimates — models tokenize punctuation and code differently.

Rule of thumb: about 1 token per 4 characters, or 1.3 tokens per English word. Code, CJK text, and unusual punctuation can differ a lot.

This is an estimate for planning, not a vendor tokenizer.
If the 128k row is already full, split the job or retrieve instead of pasting the whole corpus.
Practice the mental model in What is an LLM, then pack a context window on purpose.