Home / Token counter

Free token counter for ChatGPT and Claude

Paste any text to estimate how many tokens it is and how much of each context window it would use. It runs entirely in your browser; your text is never sent anywhere.

How the estimate works

A token is a chunk of text a model reads as one unit: a short word, part of a longer word, a number or a piece of punctuation. In English it averages about four characters. This counter divides your character count by 3.9 for the middle figure and shows a range from 3.4 to 4.4 characters per token, which covers typical prose and code.

It is an estimate on purpose. Each model family has its own tokenizer, and Anthropic has not published one for Claude, so any in-browser Claude count is approximate. For checking whether a prompt fits a context window, that is accurate enough.

Tokens and context windows

Your prompt, the whole conversation so far and the reply must all fit inside the model's context window. The bars above show what share of each window your text alone would use. The window sizes are the defaults our extension uses; sources disagree, so see context limits by plan.

Want it to count your live chats instead?

Pasting text only measures a prompt. The ContextMeter extension measures the whole conversation as you chat, on ChatGPT and Claude, and warns you when it is time for a fresh chat. See how it works.

Frequently asked questions

How many tokens is a word?

In English, a token is about three quarters of a word, so 100 words is roughly 130 tokens. Code, numbers and non-English text usually use more tokens per word.

How many tokens is a page?

A page of about 500 words is roughly 650 to 700 tokens. Treat it as a rule of thumb, since formatting and language change it.

Is this token counter exact?

No. It estimates from the character count and shows a range. Exact counts need the model's own tokenizer, and Anthropic does not publish one for Claude. For sizing a prompt against a context window an estimate is usually enough.

Is my text sent anywhere?

No. The counting runs in your browser. The page makes no requests with your text and stores nothing.

Why do ChatGPT and Claude count differently?

Each model family splits text into tokens in its own way, so the same text can have a different token count on each. Claude tends to use a few percent more tokens than GPT models for the same English text.

Last reviewed 2026-10-08.