Free tool
AI readability checker
Paste any text to get five readability scores at once — Flesch Reading Ease, Flesch–Kincaid, Gunning Fog, SMOG and Coleman–Liau — plus the sentences, passive constructions and filler phrases that are driving those numbers.
It also shows the patterns AI detectors react to, which are not the same thing as readability and sometimes point the other way. Free, no signup, and nothing leaves your browser.
Your text
Flesch Reading Ease
—/ 100
Paste your text to see scores
- 7 readability formulas, with a median grade level
- Long and passive sentences highlighted as you type
- Filler adverbs and wordy phrases, with shorter versions
- The patterns AI detectors react to
Grade levels need about 100 words to settle — below that a single long sentence moves them by several grades, so the tool waits rather than showing a number it would have to walk back.
What each readability formula measures
Every formula here is a proxy. None of them reads your argument or knows your subject — they count structural features and infer difficulty from them. Knowing which feature each one counts tells you which number to trust for your kind of writing.
Flesch Reading Ease — the 0–100 headline
Combines average sentence length with syllables per word, returning 0–100 where higher is easier. Above 60 is comfortable general reading; below 30 is dense professional prose. It is the most stable score on short passages, which is why this tool leads with it.
Flesch–Kincaid Grade Level — the default in SEO tools
The same inputs as Reading Ease, expressed as a US school grade. If a content brief specifies a reading level without naming a formula, this is almost always the one it means.
Gunning Fog — long words as the difficulty signal
Counts words of three or more syllables alongside sentence length. Developed for business writing, it is unusually sensitive to jargon, which makes it useful for spotting text that is technically simple but terminology-heavy.
SMOG — the healthcare standard
Built for material where being misunderstood carries real consequences, and widely used for patient information. It needs at least three sentences to return a result. Where SMOG and Flesch–Kincaid disagree sharply, polysyllabic vocabulary is usually the cause.
Coleman–Liau and the Automated Readability Index
Both count characters rather than syllables, so they sidestep the syllable-guessing errors that trip up other formulas on names, acronyms and technical terms. When your text is full of unusual words, these two are the more reliable pair.
Readable and “human-sounding” are different targets
This is the part most readability tools leave out, and it matters if you are editing with detectors in mind.
Readability formulas reward short sentences and common words. AI detectors flag text that is statistically predictable — which is largely produced by short sentences and common words. Optimise hard enough for a readability score and you can push your text toward the profile detectors treat as machine-like, while a rambling, unpredictable draft scores badly for readability and looks unmistakably human.
The property that separates them is variation. Writing that mixes a twenty-eight-word sentence with a four-word one reads well and carries the irregularity detectors associate with people. That variation is called burstiness, and it is the single measurement worth optimising for both goals at once. How burstiness works
Worth stating plainly: a good readability score is not protection against being flagged, and we are not going to imply otherwise. If you have already been accused, the AI patterns tab above shows the properties a detector reacts to in your own text — the research on how accurate those detectors actually are is the more useful thing to bring to that conversation, especially if English is not your first language.
How to improve a readability score
- 1
Split the longest sentence first. Both inputs to every formula are dominated by sentence length. Cutting one thirty-five-word sentence in half moves the score more than rewording a whole paragraph.
- 2
Replace long words only where a shorter one means the same thing. “Use” for “utilise” is free. Swapping a precise technical term for a vague common one buys you a point and costs you accuracy — a bad trade.
- 3
Convert passive clauses where the actor matters. “The committee rejected the proposal” is shorter and clearer than “the proposal was rejected by the committee.” Leave the passive where the actor genuinely is not the point.
- 4
Cut filler adverbs. “Very”, “really”, “actually” and “basically” almost never survive a careful edit. Removing them shortens sentences without touching meaning.
- 5
Then deliberately leave some variation in. Once the score is where you want it, resist flattening every remaining sentence to the same length. Uniformity is what makes edited prose read as mechanical, to human readers and detectors alike.
What these scores cannot tell you
Readability formulas were designed in the 1940s through 1970s, most of them for print material and military manuals. They count syllables, characters and sentence boundaries. That means a passage of fluent nonsense and a passage of careful argument with identical structure receive identical scores.
They also mishandle several ordinary cases: bulleted lists read as one enormous sentence unless punctuated, proper nouns and acronyms inflate syllable counts, and code samples or numbers distort every formula that counts characters. If your text contains a lot of any of these, treat the numbers as directional.
The useful way to read this page is as a diagnostic, not a target. The sentence highlights tell you where readers will struggle; the scores just summarise how often that happens.
Frequently asked questions
- What is a good readability score?
- For general web writing, aim for a Flesch Reading Ease of 60–70, which lands around a US grade 8–9 reading level. Journalism typically sits near grade 8; academic and legal writing runs to grade 15 and above. There is no universally correct target — a cardiology journal should not read at grade 8. Match the level to the audience you actually have, not to the highest score you can reach.
- What is the difference between Flesch Reading Ease and Flesch-Kincaid Grade Level?
- They use the same two inputs — sentence length and syllables per word — but report them on opposite scales. Reading Ease returns 0–100 where higher means easier. Flesch-Kincaid Grade converts the same measurement into a US school grade, where higher means harder. A Reading Ease of 60 corresponds to roughly grade 9. Reading Ease is the more stable of the two on short passages.
- Why do the five formulas give different scores for the same text?
- Because they measure different proxies for difficulty. Flesch and SMOG count syllables; Coleman-Liau and the Automated Readability Index count characters, which avoids syllable-guessing errors on unusual words; Gunning Fog counts words of three or more syllables. Text with bullet lists, proper names, or technical terms will spread these formulas apart. A spread of two to three grades is normal, and the median across all five is more trustworthy than any single number.
- Is passive voice bad writing?
- No. Passive voice is the right choice when the actor is unknown, irrelevant, or deliberately de-emphasised — 'the samples were refrigerated overnight' is better lab writing than naming who opened the fridge. It becomes a problem only when it accumulates, because each passive clause adds words and pushes the actor out of the sentence. Treat the count here as a prompt to check, not a list of errors.
- How many words does this tool need?
- About 25 words for a reading-ease score and roughly 100 for grade levels. Every grade formula divides by sentence count, so on a short passage a single long sentence can move the result by several grades. The SMOG formula needs at least three sentences before it returns anything at all. Below these thresholds the tool reports the score as unavailable rather than showing a number it would have to caveat.
- Does a better readability score mean my text will not be flagged as AI?
- No, and often the opposite. Detectors react to predictability — even sentence lengths, a constrained vocabulary, conventional phrasing. Simplifying your writing can raise a readability score while making it look more machine-like on exactly those measures. The two goals genuinely pull against each other, which is why this tool reports them in separate tabs instead of blending them into one number.
- Is my text uploaded anywhere?
- No. Every formula and highlight is computed in your browser with local JavaScript. Nothing is transmitted to a server, stored, or logged — which is also why there is no signup, no word limit, and no usage cap.