Turnitin vs Originality.ai vs Copyleaks: Pricing for Schools Compared

A teacher emails the IT office asking to “buy the AI checker,” and the reply comes back as a quote with three product lines, an enrollment count, and a multi-year term. That gap — between what people think detection costs and how it’s actually sold — trips up almost every school comparing tools. Turnitin, Originality.ai, and Copyleaks don’t just charge different amounts. They charge in completely different *shapes*, and the shape matters more than the sticker.

Key takeaways

• The three tools bill in completely different ways: Turnitin sells annual per-student institutional contracts, Originality.ai sells pay-as-you-go word credits, and Copyleaks sells credit or page plans with a separate education track.
• Turnitin almost never publishes a per-seat price. It’s negotiated, bundled with plagiarism checking, and priced per enrolled student per year.
• Originality.ai’s credit model makes cost easy to predict for a fixed volume but awkward to budget across a whole campus.
• Copyleaks splits self-serve plans from education and enterprise licensing, so the public price is rarely what a district pays.
• Whatever a school pays, the score is still a probability, not proof. Human judgment has to come last.

Three pricing models, not three prices

Here’s the single most useful thing to understand before you compare numbers: these vendors don’t compete on price the way two coffee shops on the same block do. They compete on *how* you pay.

Turnitin: per-student, per-year, quoted privately

Turnitin is the incumbent on campus, and it’s priced like one. There’s no public price list and no “add to cart.” An institution signs an annual contract, and the cost is typically calculated per enrolled student per year, regardless of how many papers each student actually submits. The AI-writing indicator isn’t a separate purchase you bolt on. It ships inside the same similarity report instructors already see, so you’re really buying the whole plagiarism-plus-AI bundle.

Figures reported by schools over the years have hovered around a few dollars per student annually, but treat that as folklore, not a quote. The real number flexes with enrollment, contract length, and whether you’re also licensing Gradescope or feedback tools. A 20,000-student university and a 400-student private school end up with very different per-seat math.

The practical consequence is awkward: a single professor, or even a small department, usually *can’t* buy Turnitin on their own. It’s an institution-level decision. That’s exactly why frustrated instructors end up running essays through free public tools instead.

Originality.ai: word credits, fully public

Originality.ai sits at the opposite end. Its pricing is published, self-serve, and built on credits — about 1 credit per 100 words scanned for AI detection (plagiarism checking and deep scans cost more per 100 words). You buy credits through a monthly subscription or a pay-as-you-go pack, and you spend them as you scan. No enrollment count, no contract, no sales call for ordinary volumes.

That transparency is genuinely handy when you’re budgeting a known quantity of text. The catch is that Originality.ai was built mainly for content teams, agencies, and publishers, not for an LMS gradebook. There’s no native “every Canvas submission gets scanned automatically” workflow the way Turnitin offers. For a school, that means either manual copy-paste checking or building something with their API. Either way, that staff time is a hidden line item the credit price never shows.

Copyleaks: two tracks wearing one logo

Copyleaks is the shape-shifter. Visit the pricing page and you’ll see self-serve plans measured in credits or pages, with a credit covering a set number of words and overage options on top. That’s the public face.

But Copyleaks also runs a distinct education and enterprise track: LMS integrations for Canvas, Moodle, and Blackboard, admin dashboards, single sign-on, and volume licensing, all quoted directly. So the price you read online is the self-serve number, while the price a district actually pays for campus-wide deployment is a custom figure you only get by asking. If you’ve ever heard two schools quote wildly different Copyleaks costs, this is usually the reason.

A mini-scenario: same need, three bills

Picture Maple Ridge Community College: 6,000 students, wanting AI detection across all its writing-intensive courses.

• Turnitin quotes an annual per-student contract bundling plagiarism and the AI indicator, wired into the existing Canvas setup. One number, one renewal, every submission covered automatically. Predictable, but you’re paying for every enrolled student whether they write one essay or twenty.
• Originality.ai would, in theory, be cheaper per scan. But covering 6,000 students’ submissions means estimating total word volume across a term, and that’s a guess. Get it wrong and credits either run out mid-semester or sit unused. Someone still has to wire it into the workflow, too.
• Copyleaks routes Maple Ridge to its education track for an integrated quote that competes with Turnitin on features, landing somewhere the public credit plans never advertise.

Same requirement, three very different conversations. And notice that only one of them — Originality.ai — lets a department head sanity-check the cost without a sales call.

What actually drives the bill

Strip away the marketing and a handful of levers move every quote:

• Headcount vs. volume. Turnitin charges for *who* could submit. Originality.ai charges for *how much* you scan. Copyleaks does a bit of both, depending on the track. A school with many students who each write a little is cheapest on a volume model. A school with few students writing constantly is cheapest on a flat per-seat model.
• Integration. Native LMS plumbing is the expensive, valuable part. It’s also why the cheap public price and the real institutional price drift so far apart. You’re paying for the integration, not the algorithm.
• The bundle. Turnitin’s AI score is inseparable from its plagiarism product. If you already want similarity checking, the marginal cost of AI detection is close to nothing. If you *only* want AI detection, you’re paying for features you won’t touch, and a focused tool may fit better.
• Contract term. Multi-year commitments lower the per-student rate. That’s standard enterprise software, and it’s worth modeling before you sign anything.

For a side-by-side of how a focused AI-detection plan is priced when it’s *not* buried in an enterprise contract, you can compare plan pricing and see the credit-style model out in the open.

The reliability footnote nobody puts on the invoice

Here’s the part that belongs at the top of every procurement doc: paying more does not buy a more trustworthy score. All three tools output a probability that text is AI-generated, never a verdict. They can flag real human writing, and they do.

This isn’t a fringe worry. A 2023 Stanford study (Liang et al., published in *Patterns*) found AI detectors disproportionately misclassified writing by non-native English speakers as machine-generated — in their tests, well over half of the non-native samples got wrongly flagged. OpenAI quietly retired its own AI classifier in July 2023 because its accuracy was too low to be useful. The signals these tools react to — how predictable your word choices are, how much your sentence rhythm varies — are statistical, and they wobble. If you want the mechanics, here’s the statistics behind how detectors actually score text.

A pricey institutional license buys integration, logging, support, and a paper trail. What it doesn’t buy is certainty. Whatever the school pays, a human still has to read the flagged work, look at drafts and version history, and treat the number as one input rather than a sentence. If you want to dig into how different detectors compare on the things that aren’t price, you can browse more detector breakdowns.

Frequently asked questions

How much does Turnitin cost for a school?

Turnitin doesn’t publish a public price. It’s an annual institutional contract, typically priced per enrolled student per year and quoted after a sales conversation. Reported figures land around a few dollars per student annually, but treat that as hearsay. The real number depends on enrollment, term length, and which products you bundle. A single teacher generally can’t buy a standalone seat.

Is Originality.ai cheaper than Turnitin for a school?

It depends on volume. Originality.ai sells word credits (about 1 credit per 100 words for AI detection) with no per-student contract, which can be far cheaper for a small program with predictable usage. For a large campus scanning everything, the credit math grows and the missing LMS integration costs staff time. Price it against word volume, not headcount.

Does Copyleaks have special education pricing?

Yes. Copyleaks runs public self-serve credit and page plans plus a separate education and enterprise track with LMS integration, admin controls, and volume licensing quoted directly. The website price is the self-serve number. Integrated district deployments get a custom quote.

Why won’t these companies just publish one price?

Because institutional software sells on negotiated contracts, not shelf prices. Enrollment, term, bundled products, and integration all move the number. Originality.ai is the exception, with fully public self-serve pricing, though even it offers custom enterprise deals at high volume.

If a school pays for a detector, does that make its score reliable?

No. Price and accuracy aren’t connected. Every tool outputs a probability, not a verdict, and all of them can flag genuine human writing. A paid license buys integration and support, not certainty. The score still needs human review before any consequence.

The bottom line

The cheapest detector on paper isn’t always the cheapest in practice, and the priciest contract doesn’t buy a more honest answer. Match the *pricing shape* to your school — per-student for big, integrated campuses, credits for small, predictable programs — and never let the number on the invoice stand in for human judgment.

Want to see what a detector actually reports on a real piece of writing before any of this matters? Run a free AI-detection check and read the score the way a thoughtful instructor should: as a signal, not a sentence.