Paste any text to see where a rough token split falls, color-coded token by token, with a live token count and character count. Nothing leaves this page.
Heads up: this is an approximation, not any specific model's exact tokenizer. It uses a GPT-style regular expression that splits on word boundaries, contractions, numbers, punctuation, and runs of whitespace. Real tokenizers (BPE, SentencePiece, tiktoken, and so on) merge and split sub-words differently, so treat the count as a ballpark and not a billing figure.
All tokenizing happens locally. Your text is never sent to any server.