GPTKit
GPTKit is an English-only AI text detector that runs six separate detection models over pasted text or uploaded files, returning an authenticity report and a claimed accuracy of around 93%. Guests analyse 2,048 characters free.
What is GPTKit?
GPTKit is a web tool that estimates whether a passage of English text was written by a person or produced by a generative model such as ChatGPT. It is published by Dreamsoft Innovations Private Limited, an Indian software studio whose main business is augmented reality, virtual reality and mobile development; the homepage currently advertises version 3.
Rather than relying on a single classifier, GPTKit combines six different AI-based detection methods and merges their outputs into one judgement. The site claims accuracy of roughly 93%, measured against a set of more than 100,000 human and ChatGPT responses, with the detector itself trained on what the FAQ calls one million datasets. The result is presented as an authenticity or reality rating accompanied by a short report, not merely a binary label. The site advises submitting more than fifty words for reliable output.
Anyone can use it without registering: the first 2,048 characters of each request are free for guests, and a free account keeps the same per-request allowance. Consumption is metered in credits, where one credit covers 1,024 characters. Text can be pasted directly or uploaded, and the upload widget accepts a wide range of formats including .docx, .xlsx, .pdf, .csv, .rtf, .epub, .odt and common image types.
Paid tiers lift the ceilings rather than adding features. Requests grow from 2,048 characters on the free plan to 10,240, 30,720 and 153,600 characters; rate limits rise from five to ten, twenty and twenty-five requests per minute; and team size runs from two to eight to twenty members, with credits pooled across everyone in a team. API access is not included on the entry-level paid plan: it begins at the Team tier, with tokens managed from the account area, and the same endpoint is also distributed through RapidAPI.
Two limits shape who the tool suits. Only English is supported, and the accuracy figure is the vendor's own, published without a protocol or third-party verification. Submitted text is described as stored temporarily for processing and deleted from the servers once detection completes.
What it does
- Detect whether a passage of English text was written by a human or generated by ChatGPT
- Score content across six independent detection methods at once
- Read an authenticity report rather than a single pass-or-fail verdict
- Analyse uploaded documents, spreadsheets, PDFs and images without retyping them
- Call the detector programmatically through the REST API
- Share a pooled credit balance across a team of up to twenty members
- Test any text for free, without creating an account
When to use GPTKit / When not to
A quick filter to help you decide if GPTKit is the right fit.
When to use GPTKit
- Teachers and lecturers who want a quick second opinion on submitted coursework
- Editors and proofreaders screening freelance copy before it goes to publication
- Content and marketing managers verifying that outsourced articles were written by people
- Freelance writers who need to show that their own drafts read as human
- Small teams sharing a pooled credit balance across two to twenty seats
When not to use GPTKit
- Anyone working outside English, which is the only language the detector supports
- Organisations requiring EU data residency, since data is transferred to and processed in India
- Institutions wanting defensible evidence for disciplinary decisions, given a self-reported 93% accuracy
- Buyers who need a signed DPA or a published subprocessor list, neither of which exists
- Users on the entry-level paid plan who need the API, which starts only at the Team tier
How to use GPTKit
A typical end-to-end flow, from setup to results.
- Open the homepage, where no account is needed for a first test
- Paste your English text into the box, or use Upload Files to send a document
- Watch the live counter showing characters used against the 2,048 guest limit and credits consumed
- Aim for more than fifty words, which the site says gives better results
- Press Submit and wait for the analysis to finish
- Read the authenticity report and the human-versus-generated rating
- Register a free account to keep the per-request allowance under your own login
- Choose Upgrade to compare the paid plans and switch between monthly and annual billing
- Create a team and invite members if credits need to be shared
- Generate an API token from the account area, or connect through RapidAPI, to automate detection
Pros & Cons
Pros
- Free to try immediately, with no account and no card details
- A permanent free plan rather than a time-limited trial
- Very low entry price for a paid tier at 5 USD per month
- Quotas, rate limits and seat counts published plan by plan
- Six detection methods combined, with a report instead of a bare verdict
- Broad file upload support, so documents do not have to be retyped
- Submitted text is stated to be deleted once detection completes
Cons
- English is the only supported language
- The 93% accuracy claim is self-reported, with no published protocol or independent check
- No postal address, about page, contact page or legal notice anywhere on the site
- No data processing agreement, no subprocessor list and no Article 27 EU representative
- Data is transferred to and processed in India with no transfer mechanism named
- API access is absent from the cheapest paid plan
- Terms and privacy policy are generic templates, unchanged since January 2023
Pricing & Plans
A permanent free plan is available and requires no account, covering 2,048 characters per request at five requests per minute. The lowest paid entry point is the Unlimited (Personal) plan at 5.00 USD per month, or 36.00 USD per year on annual billing. Paid tiers run up to 20.00 USD per month for the largest plan. Refunds are offered within thirty days of the original purchase.
- 0 USD - 2
- 048 characters per request
- 5 requests per minute
- 5 USD per month or 36 USD per year - unlimited credits
- faster processing
- 2 team members
- 10 requests per minute
- 10
- 240 characters per request
- no API access
- 10 USD per month or 72 USD per year - unlimited credits
- faster processing
- 8 team members
- 20 requests per minute
- 30
- 720 characters per request
- API access
- 20 USD per month or 144 USD per year - unlimited credits
- faster processing
- 20 team members
- 25 requests per minute
- 153
- 600 characters per request
- API access
Data, GDPR & hosting
A consolidated view of how GPTKit handles your data.
GDPR overview
Implementation is partial and template-based. The privacy policy, effective 24 January 2023 and generated with PolicyMaker.io, devotes a full section to EU and EEA residents, listing rights of access, rectification, erasure, objection, restriction, portability and withdrawal of consent, together with the right to complain to a supervisory authority. Requests go to an email address hosted on the publisher's own domain rather than the product's. Beyond that, the essentials are missing: no Article 27 EU representative is designated, no data protection officer is named, no data processing agreement is offered and no subprocessor list is published. Personal data is transferred to India and processed there with no transfer mechanism named, neither standard contractual clauses nor an adequacy decision. The document has not been revised since January 2023.
Who owns the data?
Dreamsoft Innovations Private Limited names itself the data controller for personal data collected through GPTKit, while the privacy policy states that users remain the sole owners of that data and may request disclosure or deletion at any time. The terms reserve the service's own content, features and functionality to the company and its licensors, but they claim no ownership over text submitted for analysis, and no clause transfers rights in user-supplied material to the vendor. According to the FAQ, submitted text is held only long enough to run the detection and is then removed from the servers.
Reuse rights
The terms are restrictive in one direction and silent in the other. Content published on the service belongs to Dreamsoft Innovations Private Limited or is used with permission, and may not be distributed, modified, copied or reused for commercial purposes or personal gain without prior written consent. Nothing restricts what users do with their own text or with the report produced from it, so detection results can be reused freely. One prohibition is explicit: the service may not be used to train other GPT detection software. On the vendor's side, the privacy policy lists thirteen processing purposes covering delivery and maintenance of the service, customer support, analytics, usage monitoring, billing and collection, account notices and marketing. Marketing messages can be declined through an unsubscribe link. Personal data is never sold or rented, and card details are handled solely by PCI-DSS-compliant payment processors.
Data retention & training
Hosting summary
The privacy policy names one jurisdiction and only one. Users located outside India are told that their information, including personal data, is transferred to India and processed there. No other hosting country is mentioned, no broad region such as the EU or North America is offered as an alternative, and no infrastructure provider is named. For EU and EEA residents this matters, because the policy identifies no transfer mechanism, neither standard contractual clauses nor reliance on an adequacy decision. No subprocessor list is published, so the third parties touching the data cannot be enumerated from the site. The domain resolves to a Cloudflare anycast address, which reflects the content delivery layer sitting in front of the site and says nothing about where processing actually happens. On the content itself, the FAQ states that text submitted for detection is stored only for the duration of processing and removed from the servers once detection is complete.
Things to keep in mind
Risks and trade-offs to weigh before adopting GPTKit.
- Treating a probability score as proof: a false positive can cost a student a grade or a writer a contract
- Over-relying on a single vendor-reported accuracy figure that no independent party has checked
- Running non-English text through an English-only detector and trusting the result anyway
- Pasting confidential or personal text into a third-party service whose retention is described only in general terms
- Sending EU personal data to a processor in India with no named transfer mechanism
- Automatic subscription renewal continuing unnoticed after the tool has stopped being used
- Letting automated screening replace reading the work and talking to the person who wrote it
Setup & Integrations
Technical difficulty
Very low for web use. The detector runs from the homepage with no installation, no configuration and no account: paste text or upload a file, then press submit. A free account takes only an email address, and creating a team is a matter of sending invitations. API use is slightly more involved but still light, requiring a token generated from the account area, or a subscription through RapidAPI instead. No SDK, connector or integration has to be configured.
Deployment
Integrations
Supported languages
Behind GPTKit
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does GPTKit actually detect?
How does the detection work?
How accurate is it?
Can I use it for free?
Which languages are supported?
What does the cheapest paid plan cost?
Is there an API?
Is my submitted text kept?
Where is my data processed?
Is there a minimum age?
Should you pick GPTKit?
GPTKit does one job and asks very little of anyone who wants to try it. Six detection methods run behind a single score, the first 2,048 characters cost nothing and need no account, and the cheapest paid tier is 5 USD per month, well below most content-authenticity tools. Quotas are unusually transparent: characters per request, requests per minute and seats per team are published for every plan, so it is easy to work out in advance whether the tool fits a given workload.
The limits are just as clear. Only English is supported. The headline accuracy of around 93% is the vendor's own figure, published without a protocol or third-party verification, and that matters because the output is exactly the kind of evidence people are tempted to act on. A false positive on a student's or an employee's work carries a real cost, and nothing on the site suggests the score should carry that weight alone.
The corporate side is thin. GPTKit is a side product of Dreamsoft Innovations Private Limited, an Indian studio otherwise focused on AR, VR and app development. The site publishes no postal address, no about page and no contact page; the terms and privacy policy are generic templates untouched since January 2023; and personal data is transferred to India with no transfer mechanism named. Organisations with procurement or GDPR requirements will find no DPA, no subprocessor list and no Article 27 representative.
Taken for what it is, a cheap and fast second opinion on short English passages, GPTKit is reasonable value, and the free tier makes evaluating it almost costless. As the basis for a decision affecting someone's grade, job or publication, it is not enough on its own.
- Choosing a selection results in a full page refresh.
- Opens in a new window.