Toolvore

Word & Character Counter

Count words, characters, sentences, and paragraphs in your text.

This tool runs entirely in your browser. Your data is never uploaded, never stored, and never leaves your device.

This counts words, characters, sentences and paragraphs as you type, so you can hit a limit exactly — a 160-character meta description, a 500-word essay, a 280-character post.

How to use it

  1. 1Paste or type your text.
  2. 2Every count updates live as you edit.
  3. 3Trim until the number you care about is where you need it.

Example

Input
one two three four five
Output
5 words · 23 characters

Characters are counted both with and without spaces, because different limits mean different things.

What happens to your data

This tool runs entirely in your browser. Your input is never uploaded to a server, never stored, and never logged. Nothing is sent anywhere, including drafts you never publish.

Last updated August 2026

Count words, characters, sentences, and paragraphs in your text as you type. Whether you're meeting a word limit for an article, a meta description, or an assignment, instant counts remove the guesswork.

Paste or type your text and see every metric update live.

Before counting anything, work out whose limit you are meeting. A human limit — an editor asking for 800 words, a 2,000-word cap on an assignment — is measured in words, and small disagreements between counters rarely decide anything. A machine limit — a database column, a meta description, an SMS segment — is measured in characters, and there the definition of a character is the whole question: the figure on screen here is what a reader would call a character, not necessarily what the system enforcing the limit counts.

The other thing to know early: this takes plain text and nothing else. There is no file picker, so a .docx or PDF must be opened elsewhere and pasted in, and whatever was formatting rather than text is gone on arrival: a numbered list comes through as its numerals, a bulleted one as its dashes, and each of those counts as a word. If your limit excludes headings, captions or references, strip them before pasting rather than subtracting after.

How it works

Toolvore counts everything in your browser in real time — your text is never uploaded. The four structural figures come from three splits and nothing cleverer. Words are the runs of non-whitespace left after trimming, so anything with a space either side counts as one — a numeral, a lone dash, a bullet character — and an unspaced script such as Chinese arrives as a single word. Sentences are what falls between runs of full stops, question marks and exclamation marks, a run of any length counting as one break. Paragraphs are the blocks separated by a blank line, so a single line break never starts a new one. Characters are grapheme clusters rather than code units, which is why an emoji reads as one, and the no-spaces figure drops every cluster that is entirely whitespace — a non-breaking space included, though not a zero-width space, which stays invisible and counted.

Common use cases

  • Staying within word or character limits for content and SEO
  • Checking meta description and title lengths
  • Measuring reading length for articles and scripts
  • Analysing text density and structure
  • Sizing a draft against a database column or API field before it is rejected
  • Setting a word budget section by section while drafting something long

Frequently asked questions

Does it count characters with and without spaces?

Yes — it reports characters, words, sentences, and paragraphs.

Is my text private?

Yes — all counting happens locally in your browser.

Why does my word count differ from Microsoft Word or Google Docs?

Counters disagree about what a word is, and none of them is wrong. Here a word is a run of characters with whitespace on both sides, so a hyphenated compound is one word, an em dash set tight between two words joins them into one, and the same dash with a space either side becomes a word of its own. Word processors tokenise by their own rules and, depending on the setting, count material you cannot see in the body — footnotes, headers, text boxes. Where a few words either way would breach a limit, ask which counter the person enforcing it reads.

Why is the sentence count higher than the number of sentences I wrote?

The split is on full stops, question marks and exclamation marks, with no knowledge of what sits around them. Every abbreviation therefore ends a sentence: 'Dr.', 'e.g.', 'etc.'. So does a decimal — '3.14' registers as a break, as do version numbers and any domain name typed mid-line. Runs of punctuation do not multiply, so an ellipsis counts once rather than three times, and a closing fragment with no full stop is still counted. Treat the figure as a structural signal — it will tell you a paragraph is one enormous sentence — rather than as exact on text full of abbreviations, prices or URLs.

Why does it say one paragraph when I can see ten lines?

A paragraph break is a blank line, not a line break. Anything whose lines are separated by a single newline — an address, a poem, a bulleted list, most text pasted out of a PDF or an email client — comes through as one paragraph however it looks in the box. The mistake runs the other way too: some PDF extractions leave an empty line after every visual line, and that text reports as many paragraphs as it has lines. Check whether your text really does separate paragraphs with an empty line, remembering that a plain text box cannot carry the styling that held the structure elsewhere.

Do emoji and accented letters count as one character?

Here, yes — the count is of characters as a reader perceives them, each grapheme cluster counting once. Most software underneath counts something else. UTF-16 code units, the default in a browser and in many databases, make a thumbs-up two and a multi-person emoji built from joiners as many as eight. A UTF-8 byte count gives a third answer again. An accented letter typed on one machine can arrive as a base letter plus a combining mark, which some counters read as two and this one reads as one. Right for a human limit, then, and wrong for a storage limit: a column that accepts twenty characters may refuse twenty emoji.

How do I count characters for a meta description, a tweet or an SMS?

Count here for a working figure, then check what the receiving system counts, because the three do not agree. A search snippet is truncated by the width of the rendered line rather than by a character total, so a run of wide capitals is cut earlier than the same count of narrow lower-case letters. X weights characters by script and substitutes a fixed length for every link. SMS is strictest: a message holds 160 characters per segment only while every character sits in the GSM-7 set, and one curly apostrophe or emoji forces it to UCS-2 and a 70-character segment.

Is the reading time realistic, and can I use it to time a talk?

It is a fixed 200 words a minute rounded up to a whole minute, which is why a single word already shows as 1 min and why the figure steps rather than climbs. Two hundred is a convention for silent reading of ordinary prose; dense material — legal text, code, tables, a second language — runs slower, and skimming runs faster. It is not a script timer. Reading aloud is considerably slower than reading silently, and pauses, questions and slide changes add time no word count can see. To time a talk, read a section aloud against a stopwatch and use your own rate.