Character counter
Type something and the most frequent words show up here.
Type something and the most frequent words show up here.
The count updates on every keystroke, both with and without spaces, alongside the limits people usually need it for. A character here means what a person would call one, so an emoji counts as a single character however many bytes it takes. Nothing is sent anywhere; the counting happens in the page.
How to count characters
Two counts, because two different people ask
Which number you want depends on who set the limit. Social platforms, SMS and almost every form field count characters including spaces. University submissions, journals and some competition rules more often specify characters excluding spaces, and for ordinary English prose that runs about a sixth lower. Both numbers sit side by side in the headline row so there is no arithmetic to do in your head, and the gap between them is simply your whitespace: spaces, tabs and line breaks all come out.
An emoji is one character here, and that is a decision
Under the hood a string is a sequence of code units, and one visible symbol can occupy several of them. A thumbs-up with a skin tone is two code points; a flag is two; a family emoji is seven. This page counts grapheme clusters through the browser's own segmenter, so each of those is one, matching what a single press of backspace removes. Plenty of back ends count UTF-16 code units instead and see two, four or eleven for the same text. When a field rejects a bio this tool says is comfortably short, a difference in what the two ends call a character is nearly always the cause. On a browser without a segmenter, which in practice means Firefox before version 125, the count falls back to code points, and the two agree on anything in the Latin alphabet.
Where the five limit bars come from
The SMS figure is arithmetic rather than a convention. A short message body is 140 octets. The GSM 7-bit alphabet packs one character into seven bits, and 140 octets is 1,120 bits, so 160 characters fit exactly. Use one character the 7-bit alphabet does not contain, a curly quote or an em dash pasted from a word processor being the usual culprits, and the whole message switches to UCS-2 at sixteen bits a character, which is 70. The bar here is the single-message case; concatenation adds a header inside the same 140 octets and drops the parts to 153 and 67.
The two search figures are softer. Google truncates a description and a title by rendered pixel width, not by a character count, and it rewrites descriptions outright a good deal of the time, so 155 and 60 are the character lengths that usually survive rather than published ceilings. A line of capitals or a run of wide letters overruns them; a line of narrow ones does not. The 280 for a post on X is a weighted count in which characters from the CJK ranges cost two, and a link is charged at the fixed shortener length whatever its real one.
What the rest of the numbers mean
A word here is a letter or digit followed by any run of letters, digits, apostrophes and hyphens, so "state-of-the-art" is one word and so are "don't" and "2026". Sentences are split at a full stop, question mark, exclamation mark or ellipsis, which means an abbreviation such as "e.g." registers as a sentence ending; a text thick with abbreviations reads high. Reading time uses 238 words a minute, the silent-reading average from Brysbaert's 2019 meta-analysis, and speaking time uses 130. Syllables are estimated from vowel groups with a rule for silent endings, because English spelling does not determine syllable count and no counter can be exact about it.
What people use it for
- Trimming a meta description to the length Google will actually render
- Getting an SMS back under 160 so it sends as one message rather than two
- Finding the one pasted curly quote that dropped an SMS to a 70-character limit
- Checking a bio against a field limit before the form rejects it
- Counting without spaces for a submission that specifies it that way
- Working out why a field refuses text that looks short enough
- Sizing a caption before it gets cut off in a feed
- Reading back the most frequent words in a draft to see what it repeats
Questions
Both numbers are on screen at once. Social and SMS limits count spaces; university and journal submissions more often specify characters excluding spaces, which runs about a sixth lower for English prose.
Usually emoji or accented letters. This counts what you would call a character; many systems count the underlying UTF-16 code units, so a single emoji reads as two or more.
One, here. A skin-toned thumbs-up is two code points, a flag is two and a family emoji is seven, but all of them are a single grapheme cluster, and a single backspace removes each one whole.
280 for a standard post, counted with weights: characters in the CJK ranges cost two rather than one, so the same visible length goes further in English than in Japanese. A link is charged at the shortener's fixed length however long the real URL is.
A message body holds 140 octets. The GSM 7-bit alphabet stores a character in seven bits, and 1,120 bits divided by seven is exactly 160.
One character outside that 7-bit alphabet forces the whole message into UCS-2 at sixteen bits each, and 1,120 divided by sixteen is 70. A curly apostrophe or an em dash pasted from a word processor does it on its own.
The joining header lives inside the same 140 octets and costs seven septets, so parts run 153 characters in the 7-bit alphabet and 67 in UCS-2. The bar on this page is the single-message case.
The extension set: the caret, curly and square brackets, tilde, pipe, backslash and the euro sign. Each is stored as an escape plus a character, so each eats two of the 160.
It is a working figure rather than a rule. Google truncates on rendered width and rewrites descriptions often, so a line of wide capitals is cut before 155 and a narrow one survives past it.
Same reason. Titles are cut on pixel width in the result, and 60 characters of ordinary sentence case is the length that usually fits inside it.
A letter or digit, then any run of letters, digits, apostrophes and hyphens. So "state-of-the-art" is one word, "don't" is one, and "2026" is one.
Sentences end at a full stop, question mark, exclamation mark or ellipsis, and nothing tries to recognise abbreviations. A paragraph carrying "e.g." and "Dr." reads two sentences longer than it is.
238 words a minute, the silent-reading average across Brysbaert's 2019 meta-analysis of reading-rate studies. Speaking time uses 130, which is a conversational pace rather than a presenting one.
Estimated. Vowel groups with a rule for silent endings gets most English words right and predictably fails on the awkward ones, "business" and "queue" among them, because spelling does not determine syllables.
The list drops stopwords such as "the" and "of", drops anything under three letters, and shows the ten commonest of what is left.
Yes, and line breaks, both labelled as such. The table lists the thirty commonest characters, so on ordinary prose the space is the row at the top, and seeing it sit at roughly a sixth of the total is the quickest sanity check on the two headline numbers.
It counts as a character and as whitespace, so it lands in the with-spaces number and not the without-spaces one. The line count follows the number of line breaks, so a file ending in a newline reads one line high.
None is set. It runs on your device, so a whole chapter is fine; the only ceiling is your own machine.
Once the page has loaded, yes. Typing fetches nothing.
No. There is no upload and nothing is stored; close the tab and it is gone.