Word & Character Counter

No. Every count — words, characters, sentences, reading time — is computed locally in your browser. Nothing you type or paste is sent to a server, so it's safe for confidential drafts, contracts, and unpublished work.

Reading time divides the word count by 200 words per minute — a typical silent-reading pace for adults. Speaking time uses 130 words per minute, the comfortable pace of a prepared talk or voice-over. Both are estimates: dense technical text reads slower, familiar text faster.

A word is what you read; a token is what an AI model reads — a chunk of characters from the model's fixed vocabulary. Common English words are often one token each, but longer or rarer words split into several, so a text usually has more tokens than words. This counter doesn't count tokens: token counts need a ~1 MB tokenizer download, so they live in a tool of their own. For an exact o200k (GPT-5 / GPT-4o) or cl100k (GPT-4) count with a token-by-token view, use our Token Counter & Cost Calculator.

Counters disagree on edge cases: numbers, dates, em dashes, slashes, and stray punctuation can each be counted as a word or skipped, depending on the tool. This counter treats any run of letters or numbers as one word, keeps apostrophes and hyphens inside it (so "don't" and "state-of-the-art" each count once), and ignores standalone punctuation — so its count can sit a word or two off another tool's on the same text. One case where the gap is wider than a word or two: scripts written without spaces (Chinese, Japanese) group into unbroken runs rather than dictionary words, so treat their word counts as approximate — character counts stay exact for every script.

The panel lowercases every word, counts identical forms together, and lists the most-used ones with each word’s share of the total word count — that share is the keyword density. It shows the top 10 by default; the button in the panel header cycles that through 20, 50 and 100 and back. Common English function words (the, and, of, to…) are left out, because they would otherwise win every list, but they still count towards the total the share is measured against. Words are matched exactly, not stemmed, so "post" and "posts" are counted separately.

The gauge counts characters including spaces and marks a recommended post length for seven networks across six marks — Instagram and TikTok at 150, Facebook 250, X 280, LinkedIn 300, Threads 500, Telegram 1,024. Networks that share a length stack into one marker; hover it and they fan out. A network’s icon dims once your text passes its mark. X, Threads and Telegram are the platforms’ own hard limits (Telegram’s applies to a post with media — plain text posts allow 4,096); the rest are practical targets for a post that reads well and doesn’t get cut off in a feed, not those platforms’ absolute maximums, which are much higher.