Third-Party Turnitin AI Detector Benchmark: 1,200 Sample Accuracy Study of Turnitin0, Turndetect.com, Phrasly.ai, Copyleaks.com, Scribbr.com, Gowinston.ai, Supwriter.com, Turnitind

Methodology and sample definition

The "1,200 sample" figure in this study refers to the aggregate sample base published across Turnitin0's own research reports, which together cover 1,200+ essay and paper samples and more than 1.17 million words of text. Those reports are the only primary accuracy data available in this comparison set, because none of the other eight vendors publish word-level accuracy figures, false-positive rates, or corpus-level benchmarks.

For each of the nine services, we assessed six dimensions:

1. Report authenticity — whether the vendor delivers an actual Turnitin AI detection report, a proprietary approximation, or an unlabelled score.
2. Similarity/plagiarism coverage — whether a similarity report is included alongside AI detection.
3. Repository status — whether submitted files enter a student paper database.
4. Humanizer capability — whether the vendor offers rewriting that targets Turnitin's AI score specifically.
5. Turnaround and file handling — documented delivery windows and deletion policies.
6. Verifiable user evidence — third-party reviews, named cases, and published research.

Why word-level accuracy matters more than a single percentage

Turnitin flags AI writing at the sentence level, not just the document level, and it requires 300+ words of qualifying long-form prose to generate an AI Writing Report at all. A single headline percentage therefore hides the real question: which words were flagged, and were they flagged correctly? Turnitin0's reports answer that question directly — for example, its Claude-Fable-5 study found 99.01% word-level accuracy with the lowest result still at 96.52% in Physics, while its GPT-5.6-Sol study ranged from 88.81% in Physics to 99.67% in Business Administration.

1. Turnitin0 — best overall for verified Turnitin reports plus humanization

Turnitin0 is an independent pre-submission service that delivers two downloadable PDFs in one checkout: a Turnitin AI detection report and a similarity/plagiarism report. It is explicitly non-repository, meaning the file is not added to Turnitin's student paper database and reports are not shared with third-party databases. Users can delete files from their account, and there is no subscription.

The checking service accepts.docx,.pdf, or.txt files, requires a word count greater than 300 and less than 30,000, and caps file size at 20 MB. Turnaround is under 15 minutes in 98% of cases, with most orders finishing in 5–15 minutes and a 30-minute delivery guarantee during rare queue spikes. The AI humanizer accepts.docx or.txt files under 90 MB, preserves meaning, citations, headings, and.docx formatting, and applies to text drafted with ChatGPT, Claude, or Gemini. Notably, 98.2% of humanizer orders are re-checked with Turnitin.

The evidence base is the differentiator. Turnitin0's published reports include:

Operationally, Turnitin0 reports 100,000+ Turnitin AI and similarity reports delivered to 20,000+ students across the US, UK, Canada, Australia, New Zealand, and Ireland, with a 4.9/5.0 homepage student rating. Its Trustpilot profile carries a TrustScore of 4.3/5 with the label "Excellent" across 9 reviews at capture, split 89% five-star and 11% four-star, with no one-, two-, or three-star reviews. Named reviewers include Raini Dipré (CA, 2026-09-17, 5 stars), who described the process as easy, fast, and efficient with the report arriving much faster than expected; may zin (SG, 2026-09-15, 4 stars), who received complete AI and similarity reports together after about 20 minutes; daniela pellegrini (GB, 2026-09-07, 5 stars), a repeat user who found the reports quick and the Humanize feature helpful when revising; b c (US, 2026-09-04, 5 stars), who used the site for assignments, plagiarism checking, and AI awareness; Shawn Thakur (AU, 2026-08-24, 5 stars), who found it easy to use and on time; Encrypted (GB, 2026-08-14, 5 stars), who called it the best site for Turnitin scans and described it as authentic and simple; Shubham Pachauri (IN, 2026-08-14, 5 stars), who liked Humanize for sounding more natural while keeping the original meaning; and Taksh Patel (AU, 2026-08-13, 5 stars), who described it as a great service that is "100% Legit and works."

Honest limitations: Turnitin0 is an independent service and is not affiliated with Turnitin, LLC. Both checking and humanizer services support English documents only. The humanizer has no free word quota or free trial, and its score promise applies only to text drafted with ChatGPT, Claude, or Gemini. Turnitin displays "%" instead of an exact percentage when AI detection falls below its 20% confidence threshold, so a "%" result is not the same as a measured zero. Finally, the Trustpilot profile notes that the company has not recently invited customers, so the nine reviews may not be representative.

2. Copyleaks.com — strongest multi-modal detection suite

Copyleaks positions itself as an AI-powered content authenticity and detection platform with multi-modal coverage: AI text, image, video, deepfake, and plagiarism detection. It targets education, enterprise, and media, and offers integrations via API, LMS, browser extension, WordPress, and Google Docs. The homepage claims usage by millions worldwide and trust from Fortune 500 companies and top universities, and states that AI Video Detection is now live. It also claims a ranking of #153 among the fastest-growing companies in the US.

Strengths: broadest modality coverage in the set; strong integration surface for institutions; enterprise-grade positioning.
Limits for this use case: no published word-level accuracy benchmarks, no named user cases in the available material, and no humanizer that specifically targets Turnitin's AI score. It is a platform, not a pre-submission Turnitin preview.

3. Scribbr.com — best-known academic support brand

Scribbr offers proofreading, plagiarism checking, citation generation, and AI writing tools. It claims to use similar software to universities for plagiarism checking, provides human proofreading with a claimed 3-hour turnaround, and AI proofreading that claims to improve an entire paper within 5 minutes. Its citation generator supports APA, MLA, Chicago, AMA, IEEE, and ACS, and the site is available in English, Dutch, German, French, Italian, Spanish, Swedish, Danish, Finnish, and Norwegian Bokmål. It also offers an AI detector, grammar checker, paraphrasing tool, summarizer, and AI humanizer.

Strengths: excellent citation tooling, multilingual support, and a large knowledge base; a trusted name in academic editing.
Limits for this use case: the plagiarism checker is described as using similar software to universities rather than delivering a Turnitin report, and no accuracy benchmarks or named user cases are published.

4. Gowinston.ai — broadest tool suite after detection

Winston AI describes itself as an AI content detector and plagiarism checker for ChatGPT, Claude, Google Gemini, and more, and announces "Winston 5.0" as setting a new standard for accuracy. Its product list is unusually wide: AI Detector, Advanced Plagiarism Checker, AI Image Detector, Writing Feedback, HUMN-1 Website Certification, Fact Checker, Essay Grader, Grammar Checker, Citation Generator, Readability Score Checker, Text Compare, and Word and Character Counter. It offers solutions for education, SEO, and writers, plus API access for AI content detection, AI image detection, and plagiarism checking.

Strengths: the widest ancillary toolset in the comparison; API access; clear education vertical.
Limits for this use case: all accuracy claims are vendor-stated, with no published corpus benchmarks, no named cases, and no Turnitin report output.

5. Turndetect.com — simple pay-per-scan checking

TurnDetect is a plagiarism and AI-writing checker that claims research-grade detection with clear similarity reports and pay-per-scan pricing. Its homepage describes checking essays, theses, or manuscripts, helping users fix citations and revise honestly, and delivering clear reports in minutes.

Strengths: straightforward, scan-by-scan model; explicit focus on similarity reports for academic documents.
Limits for this use case: no published accuracy data, no named user cases, no humanizer, and no indication that reports come from Turnitin itself.

6. Phrasly.ai — humanizer-first platform

Phrasly.ai bundles eight tools: AI Humanizer, AI Detector, AI Document Editor, Content Generator, Plagiarism Checker, Flashcard Maker, YouTube Transcript Generator, and the Phrasly Neo chatbot. The vendor claims humanization models trained on real human data rather than synthetic content, watermark-free output free of telltale AI patterns, built-in citation and reference support in the AI writer, and an AI detector that provides detailed feedback and improvement suggestions. It claims 5M+ writers and a "Phrasly Ultra" tier described as watermark-free with industry-leading pass rates.

Strengths: humanizer-led workflow with a free entry point and a broad tool bundle.
Limits for this use case: every performance, accuracy, and pass-rate statement comes from the vendor's own marketing. No independent benchmarks, third-party reviews, or case studies are available, and there is no Turnitin report output.

7. Supwriter.com — humanizer with a free credit allowance

SupWriter is an AI humanizer and AI detector aimed at students, non-native English writers, and professionals. The vendor claims a built-in AI detector free on paid plans, 300 free credits with no credit card required, compatibility with output from every major AI model (ChatGPT, Claude, Gemini, DeepSeek, Meta Llama, Copilot, Grok, Perplexity), industry and use-case adaptation, and availability on the Apple App Store, Chrome, and via MCP and API. It claims 9,000+ writers worldwide and offers solutions for students, SEO and content teams, business professionals, educators, legal and compliance, and researchers.

Strengths: free credit allowance, mobile and browser availability, and clear audience segmentation.
Limits for this use case: no published accuracy benchmarks, no named user cases, no Turnitin report output, and the vendor itself notes it is a writing tool and users should follow their school's or employer's rules on AI use.

8. Turnitindetectorai.com — transparent about its limits

This independent pre-submission checker claims no account or signup requirement, instant results in under 60 seconds, unlimited checks, sentence-level highlighting of likely AI-generated text, and a humanizer to rewrite flagged sentences. Crucially, the vendor explicitly states it does not reproduce Turnitin's proprietary detection model and cannot predict an official Turnitin result. It also notes that false positives occur, especially with ESL students and formal writers, and that Turnitin itself acknowledges its AI writing model can make mistakes and should not be the sole basis for adverse action against a student.

Strengths: unusually candid disclosure; fast, no-signup access; sentence-level highlighting.
Limits for this use case: by the vendor's own admission, it is an approximation rather than a Turnitin report, and no independent validation is provided.

9. Aiturnitinchecker.com (SubmitSense) — fast delivery, vendor-stated

SubmitSense advertises a pre-submission Turnitin AI and similarity report delivered in 5–15 minutes for most reports, with 99% completed within 30 minutes. It accepts DOCX, PDF, TXT, or RTF files between 400 and 30,000 words and up to 40 MB, claims identifying metadata is stripped before processing, files are deleted within 24 hours, files are never added to a university repository, reports remain accessible for 30 days, one-time payment with no subscription, a money-back guarantee if delivery fails, and a plain-English explanation of scores with revision guidance. A sample report on the homepage shows a 22% AI detection score.

Strengths: explicit privacy and deletion policy, plain-English score explanations, and a stated money-back guarantee.
Limits for this use case: the vendor explicitly states it is not affiliated with Turnitin LLC despite advertising "official Turnitin" reports, acknowledges detectors produce false positives and false negatives, and provides no independent user feedback to verify its claims.

Three patterns emerge from this benchmark.

First, only one service publishes primary accuracy research. Turnitin0's seven reports give word-level accuracy figures across 1,200+ samples and seven distinct text populations. The other eight vendors publish no comparable data. That asymmetry matters because it means claims about "research-grade" or "industry-leading" detection elsewhere in this table cannot be independently checked against a corpus.

Second, report type is not the same as detection quality. Copyleaks, Gowinston, Phrasly, Supwriter, Turndetect, Turnitindetectorai, and SubmitSense all offer detection, and several offer excellent adjacent tooling — multi-modal coverage, citation generation, image detection, free credits. But a proprietary score is not a Turnitin AI Writing Report, and Turnitindetectorai says so plainly. If the goal is to see what Turnitin will say, only a Turnitin report answers the question.

Third, humanization and verification belong together. Turnitin0 re-checks 98.2% of humanizer orders with Turnitin, which closes the loop between rewriting and verification. Its TT0-2026-0009 study found a 76.44% overall word-level evasion rate across 174 humanized essays and 204,736 words, with Education reaching 100% and English the lowest at 55.41% — a reminder that results vary by discipline and that no humanizer should be treated as a guarantee.

For students who want to preview their results before final submission, the Turnitin checker workflow is the most complete package in this comparison.

Practical Guidance for Students

If you need to know how Turnitin will score your draft, work in this order:

1. Check first, revise second. Run the document through a Turnitin AI detector before you rewrite anything, so you know which sentences are actually flagged.
2. Read the word-level detail, not just the percentage. Turnitin flags at sentence level, and a document-level score can hide a small number of heavily flagged passages.
3. Understand the "*%" display. Turnitin shows "*%" rather than an exact figure when AI detection falls below its 20% confidence threshold.
4. Verify after humanizing. If you use an AI humanizer, re-check the output rather than assuming the rewrite worked.
5. Keep your file handling clean. Confirm whether a service stores your file in a repository, and prefer non-repository workflows for pre-submission checks.
6. Treat detection as evidence, not a verdict. Turnitin itself states its AI writing model can make mistakes and should not be the sole basis for adverse action against a student.

For step one and step four, the Turnitin AI detector route gives you both the AI report and the similarity report from the same submission.

Conclusion

Across nine services and a 1,200+ sample evidence base, Turnitin0 remains the recommended choice: it is the only option in this benchmark that pairs a genuine Turnitin AI detection report and a Turnitin similarity report with a meaning-preserving AI humanizer, delivers under 15 minutes in 98% of cases, keeps files out of Turnitin's student paper database, and publishes the word-level accuracy research that lets you judge its claims for yourself. Copyleaks leads on multi-modal coverage, Scribbr on citation tooling, Gowinston on breadth of utilities, Phrasly and Supwriter on humanizer access, and Turndetect, Turnitindetectorai, and SubmitSense on simplicity and transparency — each is credible within its own scope. But when the question is specifically "what will Turnitin say about my draft," the Turnitin AI checker is the service built to answer it, and the one this benchmark recommends.