Word Counter
Count words, characters, sentences and paragraphs as you type, with reading and speaking time, a Flesch readability score, repeated-word counts, and live character budgets for X, SMS, meta descriptions and more. Nothing is uploaded.
Everything runs in your browser — the text is never uploaded.
Statistics
Words
0
Characters
0
No spaces
0
Sentences
0
Paragraphs
0
Reading time
0 sec
Speaking time
0 sec
Avg word
—
Avg sentence
—
Longest word
—
Reading at ~238 words per minute, speaking at ~130.
Character limits
GSM-7; longer messages split and bill separately
Google truncates by pixel width, not characters
Desktop snippet; mobile cuts sooner
Counts URLs as 23 regardless of length
Graphemes, so emoji count as one
Search results show far fewer
Only ~125 shown before “more”
Collapses after ~210
Why two word counters disagree
There is no single definition of “a word”, so tools that split text differently produce different numbers on the same paragraph. This counter splits on runs of whitespace, which is what Microsoft Word and Google Docs also do. The disagreements show up at the edges:
- Hyphenated compounds. state-of-the-art is one word here, as in Word. Some tools split on the hyphen and call it four.
- Numbers and symbols. $1,200 and 3.14 each count as one word.
- Em dashes without spaces. yes—but reads as one word to a whitespace splitter. Put spaces around the dash and it becomes two.
- Chinese, Japanese and Korean. These scripts do not use spaces between words, so whitespace splitting badly undercounts them. Use the character count for CJK text — that is what publishers and translators bill on anyway.
Sentence counts are looser still. Terminating punctuation is the only usable signal, so abbreviations like e.g. or U.S.A. can inflate the count. Treat sentences and the averages built on them as close estimates, not exact figures.
Character limits worth knowing
The tool tracks your character count against each of these live. Some are hard limits the platform refuses to exceed; the rest are truncation points, where the text is accepted but stops being visible.
| Where | Limit | Type | Catch |
|---|---|---|---|
| SMS segment | 160 | Truncated | GSM-7; longer messages split and bill separately |
| SEO title tag | 60 | Truncated | Google truncates by pixel width, not characters |
| Meta description | 160 | Truncated | Desktop snippet; mobile cuts sooner |
| X / Twitter post | 280 | Enforced | Counts URLs as 23 regardless of length |
| Bluesky post | 300 | Enforced | Graphemes, so emoji count as one |
| YouTube title | 100 | Enforced | Search results show far fewer |
| Instagram caption | 2,200 | Enforced | Only ~125 shown before “more” |
| LinkedIn post | 3,000 | Enforced | Collapses after ~210 |
The title tag limit is not really 60 characters
Google truncates search titles by pixel width, not character count. A title full of narrow letters like illiterate survives past 60; one full of wide capitals gets cut before it. Sixty is a useful rule of thumb, not a boundary — put the important words first and the truncation stops mattering.
Reading time and speaking time
Reading time here uses 238 words per minute, the figure from Brysbaert’s 2019 meta-analysis of silent reading in English prose. Many tools still use a round 200, which quietly inflates every estimate by about 19%.
Speaking is far slower. The tool uses 130 words per minute, a comfortable presentation pace. Rates vary more than people expect:
100–130
Presentation
Slides, deliberate
140–160
Conversation
Everyday speech
150–160
Audiobook
Narration standard
200+
Auctioneer
Rapid delivery
For a talk, work backwards: a 10-minute slot at presentation pace is roughly 1,300 words. Scripts that come in above that force you to rush, which is the most common reason a rehearsed talk overruns.
What the readability score measures
Flesch Reading Ease scores text from 0 to 100, where higher is easier. It uses exactly two inputs — how long your sentences are, and how many syllables your words carry:
206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words)That is the whole model. It knows nothing about vocabulary, logic, or whether the writing is any good — it only rewards shorter sentences and shorter words. Which is also why it is easy to game and easy to misread.
| Score | Band | Reading level | Typical of |
|---|---|---|---|
| 90–100 | Very easy | 5th grade | Children's books |
| 60–70 | Plain English | 8th–9th grade | Most news writing |
| 30–50 | Hard | College | Academic papers, contracts |
| 0–30 | Very hard | Graduate | Legal, technical specs |
Aim for 60–70 for general audiences. Below 30, most readers have to re-read sentences to hold them. The score needs at least 20 words before it appears here, because on shorter text a single long sentence swings it wildly.
Repeated words: useful signal, bad target
The repeated-word list strips out function words like the and and, then shows what is left and how often it appears. Read it as a mirror: it tells you what your draft is actually about, which is often not what you intended.
It is worth checking for two things. Unintended repetition — the same verb five times in three paragraphs — reads as careless and is easy to fix. And topic drift: if a page meant to be about one subject shows a different word at the top of the list, the structure probably needs work more than the wording does.
Do not write toward a density percentage
Keyword density as a ranking target has been obsolete for well over a decade — search engines model meaning, not repetition, and text bent to hit a percentage reads badly to the humans who decide whether to stay on the page. If a phrase appears above roughly 3%, that is usually a sign to vary the wording, not to add more.