Written and maintained by CASRAI Editorial Board
Last updated
If you searched “Pangram vs Turnitin,” you are probably trying to answer one practical question: does your institution need both, or can one replace the other? The short honest answer is that they are not really competing for the same job. Turnitin is a broad plagiarism-and-AI-detection suite built around a proprietary database of previously submitted student papers, licensed almost exclusively to institutions. Pangram is a narrower, AI-detection-first specialist that individuals, journals, and admissions offices can buy directly — and it has recently added a basic web-content matching feature of its own, though that is not the same thing as Turnitin’s institutional repository. Below is what each tool actually does, where their claims hold up, and the one honest tradeoff that should shape your decision.
Quick answer: is Pangram similar to Turnitin?
Only partly. Both can flag AI-generated writing, and both now offer some form of text-similarity checking. But they were built for different jobs and different buyers:
- Turnitin is a plagiarism-detection platform first, with AI-writing detection added as a bundled feature in 2023. It is licensed to institutions, not individuals, and its core value is a similarity index built against a proprietary cross-institutional archive of previously submitted student papers, plus web and licensed-publication content.
- Pangram was built specifically as an AI-text classifier, and remains an AI-detection specialist first. It is purchasable directly by an individual, a single department, or a journal — no institutional license required — and it has added a plagiarism-style feature that checks submitted text against public web pages, books, and news articles.
The distinction that actually matters for a compliance decision: Pangram’s plagiarism-style check, where it exists, matches against public content on the open web. It does not have access to Turnitin’s decades-deep repository of papers previously submitted by other students at other institutions — the specific database that catches a paper recycled from a past semester or copied from a classmate at another school. That is Turnitin’s real moat, and nothing in Pangram’s current product replicates it.
Tip: try code CASRAI at checkout for 15% off, if the offer is currently active for this program — codes vary by vendor and aren’t guaranteed.
What Turnitin actually does
Turnitin’s core product is a similarity/originality report: it compares a submission against current and archived web content, previously submitted student papers, and licensed periodicals and publications, and returns a similarity percentage plus a source-by-source match breakdown. That percentage is explicitly not a plagiarism verdict — Turnitin’s own guidance says a high score can reflect properly cited quotations just as easily as copied text, and reading the match report is the actual review step, not the number. Turnitin owns iThenticate, a related product built for publishers and researchers rather than classroom instructors; see our iThenticate vs. Turnitin comparison if you’re choosing between those two specifically.
Turnitin added AI-writing detection in 2023, layered on top of the same submission workflow. Turnitin’s own published figures claim roughly 98% accuracy with a false-positive rate under 1%, specifically for documents where more than 20% of the text is estimated to be AI-generated — below that threshold, Turnitin does not display a specific percentage at all, only a general indicator. Independent testing has not consistently reproduced the vendor’s false-positive figure: several published evaluations found false-positive rates in the 5–12% range, concentrated disproportionately on non-native-English writing. That gap between vendor claim and independent testing is well documented enough that Vanderbilt University disabled Turnitin’s AI-writing detection outright, and Curtin University announced it would do the same from January 1, 2026, both citing reliability concerns rather than cost. See our guide on what a Turnitin AI score actually means for the full breakdown of what the number can and cannot support as evidence.
Critically, Turnitin’s AI detection is licensed only as part of an institutional Turnitin subscription. There is no individual-purchase option — Turnitin’s own help center confirms it directly. If you are not affiliated with a subscribing institution, you cannot buy access at any price.
What Pangram actually does
Pangram was built specifically for AI-text classification, and that remains its center of gravity. On its own review page, Pangram claims 99.98% accuracy and a false-positive rate of roughly 1 in 10,000 — a figure it backs with third-party evaluation from researchers at the University of Chicago and the University of Maryland, which is a meaningfully stronger evidentiary basis than a vendor-only benchmark. A separate, publicly reported case — the NeurIPS 2026 Position Paper Track’s use of Pangram to screen roughly 970 submissions — found external audits citing a false-positive rate under 0.1% in that setting, though that figure is reported rather than independently peer-reviewed on its own terms. For the full accuracy picture, including where Pangram’s claims still deserve scrutiny, see our standalone Pangram review.
Unlike Turnitin, Pangram sells directly to individuals: a free tier (2,000 words/day, no credit card), an Individual plan at $20/month (300,000 words/month), a Professional plan at $65/month (1.5 million words/month plus API credits), and Team/Enterprise tiers for larger groups — alongside an institutional/LMS-integrated option for schools that want it. That direct-purchase model is a real structural difference from Turnitin, not a marketing detail: a single journal editor, thesis committee, or admissions office can start using Pangram today without going through a procurement cycle.
As of August 2026, Pangram has also added a plagiarism-style feature to its paid tiers, described on its own site as comparing submitted text “against billions of webpages, books, news articles, and more to identify matches with external content.” That is a real similarity check, and it is a genuine expansion beyond pure AI detection — but it is Pangram’s own claim, not yet independently audited the way its AI-detection accuracy has been, and it draws on public web content rather than a private repository of other students’ past submissions. Treat it as a useful secondary signal, not as parity with Turnitin’s institutional archive.
Pangram vs. Turnitin at a glance
| Dimension | Pangram | Turnitin |
|---|---|---|
| Core product | AI-text detection first; web-content similarity added 2026 | Plagiarism/similarity detection first; AI detection added 2023 |
| Similarity database | Public web pages, books, news articles | Web content, licensed publications, plus a proprietary archive of previously submitted student papers |
| Who can buy it | Individuals, teams, institutions | Institutions only — no individual license |
| AI-detection accuracy claim | 99.98%; ~1-in-10,000 false-positive rate (vendor claim, third-party evaluated) | ~98%; under 1% false-positive rate above a 20% AI-text threshold (vendor claim) |
| Independent testing | External academic evaluations cited by Pangram; not fully independent of vendor promotion | Multiple independent evaluations found 5–12% false-positive rates, prompting Vanderbilt and Curtin to disable the feature |
| Starting price | Free (2,000 words/day); $20/mo Individual | Not publicly listed; institutional contract only |
Figures verified against vendor sources and third-party reporting as of August 2026; re-check before relying on them for a procurement decision, since both AI-detection accuracy claims and pricing change without much public notice.
The honest tradeoff: Pangram is an upgrade layer, not a replacement
Here is the part that matters most if you are actually making a purchasing decision. An institution already running Turnitin cannot drop it in favor of Pangram and expect the same coverage. Turnitin’s similarity index draws on a repository of previously submitted student work across thousands of participating institutions built up over more than two decades — that is what catches a paper recycled from a previous term, bought from an essay mill with a paper-trail of prior sales, or copied from a student at a different school. Pangram’s own similarity feature, however capable it may turn out to be, checks against public web content, not that private cross-institutional archive. If your integrity process leans on catching recycled or purchased papers, that capability currently lives only with Turnitin (or a comparable product like iThenticate).
Where Pangram genuinely earns a look is as an AI-detection upgrade layered alongside Turnitin’s existing similarity workflow, or standing alone for a use case Turnitin does not serve well: an individual reviewer, a journal without an institutional Turnitin contract, or an admissions office that needs same-day access without a procurement cycle. If Turnitin’s AI-writing detection is the specific piece your institution has lost confidence in — and Vanderbilt’s and Curtin’s decisions suggest that is a live concern, not a fringe one — Pangram is worth evaluating on its own AI-detection merits, not as a Turnitin substitute.
Which one should you actually use?
- You run an institutional integrity program and need to catch recycled/purchased papers: keep Turnitin (or iThenticate) — nothing else currently replicates its student-paper repository.
- Your institution has lost confidence in Turnitin’s AI-writing score specifically: evaluate Pangram as a second opinion or a direct replacement for that one feature, not the whole platform.
- You are an individual reviewer, journal editor, or small program without an institutional Turnitin contract: Pangram is directly purchasable today; Turnitin is not an option for you at any price without an institutional license.
- You need both a proprietary-repository similarity check and the strongest available AI-detection accuracy: realistically, that means running both tools rather than choosing one — see our broader comparison of Turnitin’s AI detection against standalone AI detectors for how other institutions have approached that combination.
Whichever you use, treat the output the same way: a detection score, similarity percentage, or AI-writing indicator is evidence to start a conversation, not a finding on its own. Our guide on AI-detection accuracy in higher education covers what a fair review process actually requires once a score comes back elevated.
Frequently asked questions
Is Pangram similar to Turnitin?
Partly. Both detect AI-generated writing, and both now offer some text-similarity matching. But Pangram remains an AI-detection specialist purchasable by anyone, while Turnitin is a broader institution-only suite whose core strength is a proprietary archive of previously submitted student papers that Pangram does not have access to.
Can Pangram replace Turnitin at my institution?
Not fully. If your process depends on catching papers recycled from past semesters or purchased from essay mills with a paper trail, that depends on Turnitin’s (or iThenticate’s) cross-institutional repository, which Pangram’s web-content matching does not replicate. Pangram is better understood as an addition or an AI-detection upgrade than a full substitute.
Which has a lower false-positive rate?
On paper, Pangram’s vendor-claimed rate (roughly 1 in 10,000, with some third-party academic evaluation behind it) is markedly lower than Turnitin’s vendor-claimed rate (under 1%, applying only above a 20% AI-text threshold). Independent testing of Turnitin specifically has found materially higher false-positive rates in practice, 5–12% in some evaluations, which is why several universities have disabled the feature. Treat both vendors’ own numbers as a starting point, not a settled fact, and pilot against your own real documents before trusting either at scale.
Do I need my institution to buy Pangram, the way I do with Turnitin?
No. Pangram sells directly to individuals and small teams — a free tier and a $20/month Individual plan exist specifically so a single reviewer can start without procurement. Turnitin has no individual-purchase path at all; it is licensed to institutions only.
Does Pangram check for traditional plagiarism, not just AI writing?
As of August 2026, yes, in its paid tiers — Pangram added a feature that compares submitted text against public web pages, books, and news articles. That is a real similarity check, but it is not the same database Turnitin uses; it does not include other students’ previously submitted coursework, which is Turnitin’s specific value for catching recycled or shared papers.
What if Pangram and Turnitin disagree on the same paper?
Disagreement is expected and not, by itself, evidence that either tool is broken — the two products use different models and different reference data. Use a disagreement as a reason to read the actual report (which passages, which sources, which reasoning) rather than trusting either headline percentage, and follow your institution’s standard review process for a flagged case either way.
