19 people use GPTRevealer
Before you send that paper, article or pitch on, look at it through the detector's eyes. You will see exactly what the person on the other side will see. The sentences the detector reads as AI are colour-marked right in the text.
Free to try · No card required · Cancel your subscription any time with one click
Everything you need to know about a text, on one screen.
Paste a text and see the detector's score along with the sentences it marked as AI. Five detections per 30 days and texts up to 3,000 characters, no sign-in needed. Seven detectors, sentences colour-marked by each detector and texts up to 300,000 characters unlock after .
Six more detectors
A recording of the app on the Pro plan. The dashboard, detection, humanisation, questions and text history exactly as a signed-in user sees them. You can try it for free and cancel any time with one click. The plan then keeps running until the end of the paid period.
Start free and move up only once it actually starts to make sense for you.
For trying it out and the occasional check
€0/ 30 days
Cancel any time
For everyday writing and checking
€11/ 30 days
Cancel any time
For schools, newsrooms and agencies
Let's talk
Depending on team size
It is not just a different model. It differs in whose writing it was built for, what it does with your history and how it behaves when it is not sure.
Detectors often shoot from the hip and turn an ordinary complex sentence into a suspect. Our model works out the language on its own and accounts for how a sentence is built in English.
Some models leave invisible characters and marks in the text that carry over when it is copied. We look for them separately, so they turn up even in a text that otherwise reads perfectly naturally.
No plugin, no browser extension, no waiting in a queue. It works on a phone too, ten minutes before the deadline.
The models sometimes disagree, and that in itself means something. Instead of a hard verdict you see the spread and decide for yourself.
The model also looks at what you checked earlier. The more saved or uploaded texts it has, the more precisely it learns how you write, and the better both detection and humanisation fit. The free plan keeps the last seven texts, on Pro they all stay.
A teacher checks a submitted paper, an author checks their own before sending it. Both look at the same numbers, so there is something to actually talk about.
See what happens after you paste a text and press the button. Try a detection.
In today's dynamic era, artificial intelligence plays a key role. I tried it yesterday and had to rewrite half the paragraph from scratch. It is therefore important to mention several fundamental aspects.
I don't want to accuse anyone on a hunch. When something feels off, at least I have something to bring to the student — and a few times it has proved the hunch wrong.
Freelancers send texts at the last minute. Before it goes to layout I run it through and I know what I am dealing with.
A client wrote to me that it came out as AI somewhere. I wrote it myself. Since then I would rather run it through before sending.
We have a rule agreed on in the team. Six months on, the arguments about who really wrote what have stopped entirely.
I spent three months on my bachelor's thesis and the day before the deadline I wanted to go through it once more. Within seconds I knew everything was fine.
Upload thirty files at once and each one is scored separately. Open only the ones that stand out. Batch checking is part of the Pro plan.
Drag in the whole folder. No copying text one file at a time.
Results are never merged into an average. For every paper you see its numbers and the sentences the detector marked.
The suspicious papers come to the top. The rest you read for content, not for suspicion.
Three steps with nothing missing in between. From an empty account to a text whose outcome you already know.
Run a detection for freeSigning up is free and takes a minute. We don't want a card — once you confirm your e-mail you go straight into the app.
Copy in a paragraph or upload a file. Within seconds you see the score, a colour map of the text and the specific sentences the detector treats as suspicious.
If you don't like the result, send the text into humanisation. It comes back rewritten and you can measure it again right away, so you see how far the score dropped.
Something unclear? The answers to what people ask most often are all in one place here.
A free account gives you five checks and five humanisations per 30 days, texts up to 300,000 characters and 300,000 characters a month in total — whichever runs out first. No card needed. Look at a paid plan once you find it makes sense for you.
When you run out of five checks or 300,000 characters per 30 days, or when you mark work in batches — Pro takes the files at once and each one gets its own score. For one check a month a free account is enough and paying makes no sense.
The Pro plan belongs to one account and login details are not meant to be shared. For several people there is the Team plan — a shared overview, billing to a single account and onboarding at the start. The price depends on the size of the team, so write to us and we will work it out.
Yes, with one button in the menu, no calling or writing to support. Your Pro plan then keeps running until the end of the paid period and nothing more is charged. Until then you can come back to renewing with the same button at any time.
We run the text through seven detectors at once and show each one separately — we never merge them into a single number. No detector in the world is certain, which is why you also see where the results disagree.
Yes, one of the seven detectors looks for invisible characters that get into the text when it is copied out of a chat and that you cannot see in an editor. We show them separately — a zero there does not mean a human wrote the text, only that there are none in it.
Checks are saved to your history and the model draws on them so that detection and humanisation fit you better. We do not use them to train a shared model, and one button in the menu deletes your whole history along with uploaded samples.
The model is built for English and works out the language of a text on its own. If a text is in a language the model is not built for, some of the detectors have nothing to compare against, so treat the result as indicative only.
It rewrites the whole submitted text so that it has a different structure and rhythm. It follows the document type and your sample texts so that it sounds like you. Facts, numbers and tone stay the same; only the way they are written changes.
The minimum is 100 characters, the maximum 300,000 per text, which covers a whole thesis at once. The free plan handles 300,000 characters per 30 days, Pro 4,000,000 per month. For longer documents the result fits better if you split them by chapter.
No. No plugin, no browser extension. Open the page, paste a text and within seconds you have a result, on a phone too.
We are a small team, so an answer reaches you quickly. Most things, though, are already covered in the help.
Take the paragraph that has been on your mind and be sure about it before you send it on.