GPTinf

What GPTinf is

GPTinf is a humanizer with a suite built around it. The thing you pay for is the rewriter: you paste AI-generated text, pick a rewrite mode, optionally freeze the words you do not want touched, and it returns prose intended to read as human-written. Everything else is either free or unlimited on any paid plan, which tells you where the margin is. Only the humanizer draws down a word budget, and the help centre states that explicitly: on paid plans the monthly budget is spent by humanizer words processed, and the detector, grammar checker and plagiarism checker do not touch it.

The rewriter runs eight modes. The help centre names all of them, Standard, Academic, Simple, Formal, Informal, Expand, Shorten and Creative, maps each to a use case, and says mode choice is not plan-gated, so a Lite subscriber gets the same eight a Unlimited subscriber does. Freeze keywords locks names, citations and brand terms so the rewrite works around them, and it is the one feature a free account unlocks over an anonymous session. Pro and above add selective rephrase, which re-runs only a highlighted section, and re-humanize, which iterates without spending more budget. Input can be pasted or uploaded as PDF, Word or plain text on every tier, and a File-to-File Humanizer returns a whole document, capped at 5,000 words per file on Lite, 25,000 on Pro and uncapped on Unlimited. The vendor says the engine supports 11 languages.

The distinctive architectural claim is that the humanizer is not a language model. The product page says the engine edits your text directly, never passing it through an LLM; the about page calls this a custom non-AI algorithm. We cannot verify that from outside, but it is at least falsifiable, and it matches the vendor's own before-and-after examples, which restructure clauses rather than swap synonyms.

The detector is the other half, and it is the half most people will actually use. It is free, needs no account, has no word cap, labels every sentence as AI, Mixed or Human with an explanation, and returns a percentage breakdown across ChatGPT, Gemini, Claude, DeepSeek and xAI Grok. The marketing presents it as GPTinf's own work. The help centre's privacy overview describes something different: your text is sent to eight underlying detector services for scoring, over what it calls a standard API call, each with its own privacy policy. That matters in two directions. The free score is a consensus of eight outside opinions rather than one vendor's, which is more useful than a single number. It also means the text you paste into the free tool leaves GPTinf, going to eight companies it does not name.

Which brings you to the problem this product shares with every tool in the category. What nobody can verify, us included, is the part you are actually buying. Evasion is a moving target: a success rate against Turnitin, GPTZero or Originality.ai is a snapshot that expires the next time those vendors ship a model, and searching for independent humanizer testing returns mostly affiliate pages run by competing humanizers. Every detector claim here is vendor-stated rather than independently confirmed, and a single-date result would not settle it either.

Pricing

GPTinf publishes every price without a login, which puts it ahead of a good part of this category. There are three paid plans and a real free tier. The entry plan is $9.99 a month for 5,000 humanizer words on monthly billing, or $4.99 a month on annual billing, charged as $59.99 a year. Read that second figure carefully, because the pricing page makes it easy to misread: the billing toggle loads with Yearly already selected, the discount pill reads Save 50%, and the price under each plan name carries the suffix /monthly even in annual mode. The number you see first is not the number you pay first.

All figures in US dollars, read from gptinf.com/pricing in September 2026 in both billing modes. Annual totals are not printed on the pricing page; they come from the help centre's Annual vs. monthly billing article. The Terms of Service state that all payments shall be in US dollars, and that sales tax will be added where required.

How to read the toggle, the free tier and the word meter

The pricing page opens on annual and never shows you the annual charge. In its default state the Lite card reads $4.99 and the word directly beneath it is /monthly. There is no annual total anywhere on that page: no yearly figure, no billed-annually line, no per-year note. A buyer who takes the card at face value expects a $4.99 charge and gets a $59.99 one. The totals are published, clearly, in the help centre, which most buyers will never open. Clicking Monthly reveals the real monthly rates of $9.99, $24.99 and $59.99, and the annual discount is genuine at almost exactly half, but the default view is the one that misleads.

The free tier is two different things, and only one is generous. The AI detector is free and unlimited for everyone: no registration, no paywall, no word limit, on every tier including anonymous. That is not a funnel, it is the product, and it beats what several paid detectors offer. The humanizer is the opposite: 120 words per run signed out, or 240 words in total on a free registered account, splittable across runs. Two hundred and forty words is about half a page, once. Enough to see the interface and watch a score move; not enough to judge output quality on anything you would submit. There is no separate trial of the paid plans, and the help centre says so directly: the free tier is the trial.

Budget in words, and assume they expire. The monthly allowance refreshes at the start of each billing month and the help centre states plainly that unused words do not carry forward. Annual buyers get the same monthly allowance twelve times rather than a year of words up front, which the vendor flags as the thing that surprises people most. Two figures matter when you size a plan. Five thousand words a month on Lite is roughly one long article or a handful of short ones. And re-running is not free on the lower tiers: the mode-selection article warns that each run uses word budget on Free and Lite plans, and re-humanize, the feature that lets you iterate without burning words, starts at Pro. On a tool people re-run repeatedly to chase a score, that is the line that decides your real bill.

The refund window is three days, on annual plans too. The Terms offer a refund of the last payment made within three days of the date of payment, with a decision promised within five business days and processing within seven. The help centre repeats the three-day window, says refunds are reviewed case by case, and says most decisions land within one to two business days, a faster promise than the Terms make. Past three days, the vendor says it will generally stop the next renewal rather than refund the remaining term, so an annual purchase you regret in week three is a commitment you keep until the term ends. Payments run through Paddle, and the Terms list Visa, Mastercard, American Express, Discover and PayPal. One thing the pricing page says that the rest of the site does not support: the FAQ claim that all plans are month-to-month. Annual terms exist, they are the default view on that very page, and the refund policy treats them as a single charge.

What we verified, and what we did not

Reviews of tools like this often imply a score covers whether the thing works. This one does not, so here is the boundary in plain terms.

Verified: every price, plan name and word limit in the tables above, read from the pricing page in both billing modes and cross-checked against the help centre's plan, annual-billing and word-allowance articles; the free detector's no-account, no-paywall, no-word-limit status; the free humanizer caps and the absence of a paid trial; the eight rewrite modes and their plan availability; the plagiarism and grammar caps by tier; the refund window in both the Terms and the help centre; the currency and sales-tax statements in the Terms; the help centre's account of the detector calling eight third-party services; the legal operator and address; the ethics page's positions; the reviews page's transparency note; and the creator program's terms.

Not verified: whether the humanizer evades any detector, at any setting, on any date; detector accuracy; which eight services the detector calls, since the vendor does not name them; the homepage statistics, none of which carry a methodology; the 4.5-star, 5,467-review widget; the 11-language accuracy figure; whether a re-run on Pro or Unlimited deducts words a second time; uptime, latency, output quality and support responsiveness. Every item in this list is vendor-stated or simply absent from the vendor's pages rather than independently confirmed.

That means the score below is a score for pricing transparency, feature breadth, documentation and the trustworthiness of the vendor's own claims. It gives no credit, in either direction, for evasion performance.

The site's claims, side by side

Because the product's central promise cannot be tested from outside, we did the next best thing and read the vendor's pages against each other. Every pair below was live on the same day, 24 September 2026.

None of this is evidence the product is bad. It is evidence that the marketing runs ahead of what the vendor's own support documentation supports, and the marketing is the surface people buy on. The pattern runs in one direction: on every pair above, the sales page is the optimistic one and the help centre is the honest one.

Where GPTinf is the strongest pick

GPTinf is at its strongest for somebody who wants the whole cleanup loop in one tab and expects to run a lot of words through it. Paste a draft, get a detection score from eight engines with sentence-level highlighting, rewrite in the mode that fits the register, freeze the terms you cannot afford to lose, re-check, scan for plagiarism, fix the grammar, all without leaving the site or paying a second subscription. Competing humanizers sell you the rewrite and leave the rest to other people's tools.

The second argument is the ceiling. Unlimited removes the humanizer word cap entirely for $59.99 a month, or $359.99 a year. Metered competitors top out with a ladder of ever-larger word bundles; a flat uncapped plan at that price is a different shape of offer, and if your real problem is that you cannot predict your monthly volume, it is the plan in the category that solves it.

The third, and the one we would recommend to most readers, costs nothing: the free detector. No account, no cap, sentence-level output, and a consensus across eight engines rather than one. Treat the score as one input rather than a verdict and it earns its place whether or not you buy the humanizer.

Best for

Writers, marketers and agencies who are allowed to use AI, want the detect, rewrite, check and proof loop in one place, and would rather pay a flat fee than count words. Lite suits someone cleaning up one long piece a month; Pro is the sensible middle at 25,000 words; Unlimited is the reason a high-volume team would pick GPTinf over a metered rival. Also worth a bookmark purely as a free, uncapped, no-account AI detector, provided you treat its score as one input and never as a verdict. Not for students or anyone submitting graded work: the vendor's own ethics page discourages it, and institutions treat it as misconduct. Not for anyone who needs a detection result they can defend to a third party, and not for anyone who needs to automate, because we found no API and no browser extension.

Key features

• AI humanizer with eight rewrite modes, Standard, Academic, Simple, Formal, Informal, Expand, Shorten and Creative, available on every plan including free
• Freeze keywords, which locks names, citations and brand terms so the rewrite works around them; unlocked by creating a free account
• Selective rephrase, which re-runs only a highlighted section, and re-humanize, which iterates without spending more word budget, both from Pro upward per the help centre
• A detection score that runs automatically under every rewrite, so checking does not mean switching tools
• Free AI detector with no account, no paywall and no word limit on any tier, labelling each sentence AI, Mixed or Human with an explanation for the flag
• Detector model breakdown against ChatGPT, Gemini, Claude, DeepSeek and xAI Grok, built by cross-checking eight detector services the vendor does not name
• Plagiarism checker with sentence-level attribution and source URLs, 500 words a day on a free account and unlimited on every paid plan
• Grammar checker on a self-hosted LanguageTool engine, with a Fix All Errors action on paid plans, capped at 20,000 words per check
• Readability checker returning a Flesch-Kincaid level, plus a paraphraser and an essay writer metered separately from the humanizer budget
• File-to-File Humanizer, PDF, Word and plain-text upload on every tier, and 11 supported languages

Best-fit use cases

• Removing the recognisable tells from AI-drafted marketing copy, blog posts and product descriptions where AI use is permitted but robotic phrasing is not
• Running a free, uncapped second-opinion AI check on inbound freelance or agency copy, as one input alongside a detector with no stake in the result
• Rewriting human-written copy that a client's or editor's detector has false-flagged, which the ethics page names as a legitimate case
• High-volume cleanup on the Unlimited plan, where the word cap disappears and the only planning question is throughput
• Screening a draft for plagiarism and grammar in the same pass as the rewrite, instead of paying three vendors for three steps

Pros

• The AI detector is free and unlimited on every tier, with no account, no paywall and no word cap, which is more than several paid detectors give you
• That detector cross-checks eight engines and returns sentence-level labels with per-model attribution, a better-shaped answer than one vendor's single number
• Plagiarism and grammar checks go unlimited on the cheapest paid plan, so the bundle is real rather than decorative
• Unlimited at $59.99 a month, or $359.99 a year, removes the word cap entirely, a shape no metered rival in this cluster offers
• Eight rewrite modes on every plan including free, with the vendor stating explicitly that mode choice is not plan-gated, and freeze keywords available to free accounts
• The help centre is unusually good for a tool this size: billing, refunds, word budgets, per-tier limits and privacy each get a dedicated article with the numbers spelled out
• That help centre is also more honest than the sales pages, stating that no humanizer can guarantee a passing score and that detectors disagree by 50 percentage points or more
• A named legal operator with a registered address, named leadership, a claimed Trustpilot profile, and a reviews page that labels its testimonials as a curated selection and states that reviews are not incentivized

Cons

• The pricing page opens on annual billing, prints the discounted rate with a /monthly suffix, and shows no annual total anywhere, so the first number a buyer sees is not the number that will be charged
• The entry plan buys 5,000 humanizer words for $9.99 a month, half what the nearest comparable metered plan in this cluster gives for the same money
• The free humanizer allowance is roughly half a page, and the vendor publishes three different figures for it: 120 plus 120 on the pricing page, 240 in total in the help centre, and 300 per request on the humanizer page
• The pricing page FAQ claims every plan guarantees bypassing all AI detectors without exception, while the help centre and the ethics page both say no such guarantee is possible
• The detector is marketed as GPTinf's own work built on custom non-AI algorithms, while the help centre describes it as a standard API call out to eight third-party services that are never named
• Three answers to whether your text is stored: the homepage says it is not kept after the session, the about page says the company never stores user content, and the ethics and pricing pages say it is kept in a History folder
• The Lite card advertises unlimited re-humanizing and selective rephrase; the help centre puts both behind Pro, and the site gives three different counts of how many tools it sells
• Two different star ratings run on the same site, a 4.1 badge linking to Trustpilot's write-a-review form rather than the profile, and an unsourced 4.5-from-5,467-reviews widget; the Trustpilot profile itself showed 4.1 from 26 reviews
• We found no API, no developer documentation, no browser extension and no automation hook on the site or in the help centre, so everything is copy, paste and upload
• The refund window is three days from the charge, annual plans included, which is short for a product whose free tier is too small to evaluate properly, and nothing refunds against a detection outcome
• The creator program pays students $15 per TikTok plus 30% on referrals, with suggested hooks about essays and AI checkers, which sits badly against an ethics page discouraging academic use
• Whatever evasion rate the tool achieves today is perishable by design, and the homepage's four supporting statistics, including 37% better detection results than competitors, carry no methodology

What to watch out for

Four things, in order of how much they should affect your decision.

The guarantee on the pricing page is contradicted by the vendor's own support pages. This is not a nuance of wording. The pricing page FAQ states that every plan includes Ultra Humanization, which guarantees bypassing all AI detectors, without exception, and a panel below promises an engine that is always on, always 0 % AI score. The help centre, answering the same question for people who have already paid, says: does GPTinf guarantee a passing score, no, no humanizer can. It adds that a clean result across all eight of its detectors is not always achievable, and that detectors disagree with one another by 50 percentage points or more. The ethics page agrees that no service is 100% undetectable. When a vendor tells you one thing before you pay and the opposite afterward, plan around the version written for existing customers. No refund clause anywhere on the site is tied to a detection outcome.

The default pricing view is the misleading one. Load gptinf.com/pricing and the Yearly toggle is already active, the plans read $4.99, $12.49 and $29.99, and every one of those numbers sits above the word /monthly. Nowhere on that page is there a yearly total, a billed-annually note or any indication that choosing Lite means a single $59.99 charge. Combine that with a three-day refund window that applies to annual plans exactly as it does to monthly ones, and a free humanizer tier too small to evaluate the product with, and you have a buyer committing for a year on the basis of a number that was never the price. Click Monthly before you decide anything, and if you do buy annually, know that you are past the refund window in seventy-two hours.

The free detector is the best thing here and also the thing that sends your text furthest. Free, unlimited, no account, sentence-level output, eight engines cross-checked: on its own terms a genuinely good tool. But the help centre's privacy overview is explicit that it works by sending your text to eight underlying detector services, each with its own privacy policy, over what it calls a standard API call, and the vendor names none of them. The marketing says the opposite thing in spirit, presenting the detector as GPTinf's own model and calling the platform's algorithms custom and non-AI. Both can be true at once, since a cross-checking layer is still engineering, but if you are pasting confidential material, understand that the free detector is the tool that demonstrably forwards it to parties you cannot identify. The grammar checker, by contrast, runs self-hosted.

The academic framing does not survive reading the site in order. The ethics page is reasonable and reads as sincerely written: it discourages cheating and academic dishonesty, tells you to cite your sources, and asks you to follow your institution's policy. Then look at the rest of the business. Academic is one of the eight rewrite modes. The homepage's use-case grid opens with Academic Essay and Student Assignment, and the essay card says to humanize before submission and check what Turnitin will flag. The detector has dedicated landing pages for students, for college and for academic writing. The first homepage testimonial is from a high school student. And the creator program recruits students to post TikToks for $15 a video plus 30% on referrals, with vendor-supplied hooks including one about an essay that sounded like AI until the writer used this. A company can hold an ethics position and build a student acquisition engine at the same time; it just cannot expect both to be read as the same statement of intent. If you are a student, the practical point stands regardless: submitting humanized AI text is academic misconduct at essentially every institution, the vendor's own policy says do not do it, and its help centre says the tool cannot guarantee you will not be caught.

Is GPTinf worth it?

For the right buyer, yes, and the right buyer is more specific than the marketing suggests. If you are cleaning up a real volume of permitted AI copy and want detection, plagiarism and grammar in the same tab as the rewrite, the bundle beats assembling it from three vendors. If you cannot predict your monthly volume, Unlimited is the only flat uncapped offer we have found in this cluster. And if you just want a good free AI detector, take it and pay nothing.

It scores 3.3 rather than higher for reasons that have little to do with output quality, which we did not measure. Two are structural: the integration story is empty, so this is a browser tab and cannot become part of a pipeline, and the entry plan is poor value in isolation at 5,000 words for $9.99, with a free tier too small to test the product before committing to a year at a price the pricing page does not display.

The rest of the gap is self-inflicted. A vendor that writes this good a help centre, publishes a thoughtful reviews-transparency note, names its operator and address, and tells customers honestly that no humanizer can guarantee a passing score, then puts a guarantee of bypassing all AI detectors without exception on the page where money changes hands. Together, the three tool counts, three free-word figures, two star ratings and a storage answer that flips page to page mean a careful reader has to decide which GPTinf to believe, and the honest one is always the page fewer people read.

Worth saying plainly: whether the humanizer beats Turnitin, GPTZero or Pangram this month is the question most readers arrived with, and this page does not answer it. Nobody can, durably, and GPTinf's own support documentation agrees. The worst case on a monthly plan is $9.99 and three days to change your mind, so if you are going to try it, try it monthly first.

How we arrived at 3.3

The score is weighted across eight categories, each marked out of 5. The weights are published so the number can be reproduced, and value and market trust carry the most weight because they are the two things a buyer of this product can actually evaluate from outside.

The 5 point score is a direct weighted mean of the table, and it reflects a well-documented, genuinely bundled humanizer with the best free detector in its cluster, sold on a pricing page that hides the annual charge and promises a guarantee the vendor's own support pages withdraw.

FAQ

Pricing, plan limits, free-tier caps, refund terms and feature claims verified 24 September 2026 from gptinf.com/pricing (read in both billing modes), gptinf.com/humanize-text, gptinf.com/detector, gptinf.com/about-us, gptinf.com/terms_of_service, gptinf.com/reviews, gptinf.com/earn, the responsible-use policy, and the plan, annual-billing, word-allowance, refund, privacy and detection-score articles in help.gptinf.com. Every capability claim on this page is vendor-stated rather than independently confirmed, and no evasion claim is made. ToolChase received no payment for this review. Scores, rankings and list positions are never sold. See our methodology for how we score tools.