Token Counter

Tokens are the basic units that language models process. A token can be a word, part of a word, or a punctuation mark. For English text, one token is roughly 4 characters or 0.75 words on average. Different models use different tokenizer algorithms, so the same text can produce different token counts.

We support four encodings: o200k_base (GPT-4o, GPT-4.1, GPT-5, o1, o3, o4-mini), cl100k_base (GPT-4, GPT-4 Turbo, GPT-3.5 Turbo), p50k_base (text-davinci-003, Codex), and r50k_base (GPT-3). Select the encoding that matches the model you plan to use.

Our token counts use the exact same tokenizer algorithms as OpenAI. The counts are identical to what you would get from the official tiktoken library or the OpenAI API response usage field.

Cost estimates are based on publicly listed per-million-token pricing from OpenAI. We show both input and output costs since most models charge differently for each. Prices are approximate and may change — always check your provider's pricing page for current rates.

No. The tokenizer runs entirely in your browser using JavaScript. Your text never leaves your device. The tokenizer vocabulary data is bundled with the page, so no network requests are made during tokenization.

Currently we support OpenAI tokenizer encodings. Claude and LLaMA use different tokenizers. For Claude, the cl100k_base encoding gives a rough approximation. We plan to add more tokenizers in the future.