AI detector with no word limit
A word cap does not just inconvenience you. It degrades the answer, because short text is where every detector is least reliable.
Most free detectors cap the check at a few hundred or a few thousand words. That looks like a pricing decision and it is, but it has a measurement consequence that is rarely mentioned on the page where you paste the text.
Why a cap makes the result worse, not just smaller
Detectors read statistical properties, and statistics need samples.
Sentence length variation is the strongest single signal, and it is a property of a sequence of sentences. On one sentence there is nothing to vary. On three, barely anything. It takes a paragraph before the number means anything and a document before it is stable.
Predictability averages out. A single unusual sentence can be surprising or flat by chance. Over a thousand words the noise cancels.
Turnitin publishes both of its own figures, which shows the size of the effect: 4% wrong on sentences against under 1% on documents. Same tool, same model, four times worse at the smaller unit. See why detectors are unreliable on short text.
So a cap that forces you to check a 4,000-word essay in six pieces does not give you six sixths of an answer. It gives you six of the least reliable kind of answer, and no way to combine them.
Where caps hurt most
Theses and dissertations. Chapter-level is the unit where a score is both reliable and interpretable, and chapters are long. See AI detection and a thesis or dissertation.
Full papers. A methods section scored alone reads differently from the same section inside the paper, because the surrounding text is what shows you whether a flag is concentrated or spread.
Any before-and-after loop. The useful workflow is check, fix the two or three paragraphs sitting highest, check again. A per-check limit makes that impossible, and it is the part that actually changes anything.
Why ours has no cap
The honest reason is that an extra check costs us close to nothing, so there is nothing worth metering. Tools that meter do so because each check costs them money.
That is also why there is no account. Nothing to tie usage to, nothing to rate-limit, nothing to sign up for.
We do not store your text either: the only thing recorded is a word count, the resulting score, a language and a country code. See what happens to text you paste.
What you get instead of one number
Every paragraph scored separately alongside the document figure, because a single percentage over a long document is an average that hides where the signal is. In most documents two or three paragraphs carry all of it.
And the wrong-flag rate published rather than implied: 0.4%, which is 99.6% specificity, measured across 15,900 documents of real academic writing. Turnitin publishes 4%.
What it takes
.pdf, .docx and .tex directly, or pasted text, at any length. Free and unlimited, no account.