Tokenizer

Tokens, counted.

Paste text, pick a model, read the count — and see exactly how the text splits into tokens. Everything runs in your browser; each vocabulary downloads once and stays cached.