ZeroGPT vs Turnitin: the 14 fields and the three shared studies side by side
Turnitin classified 126 of 126 documents correctly in Walters 2023 while ZeroGPT classified 87% correctly with 1% incorrect and 12% uncertain, the largest gap between the two on any field in the three studies that tested both, as reported in the paper published by De Gruyter and read 8 November 2026 (Walters 2023). The same three studies put the two tools close on false accusations: 0 in 18 for both in Weber-Wulff 2023, 2% against 0 of 42 in Walters 2023, 0 in 10 for both in Perkins 2024.
Updated Fields checked 8 November 2026
Short answer
Three indexed studies tested both tools. Turnitin ranked first of 14 in Weber-Wulff 2023 with 41 of 54 documents correct against ZeroGPT's 32 of 54 (rank 6); Turnitin classified all 126 documents correctly in Walters 2023 against ZeroGPT's 87%; in Perkins 2024 Turnitin's baseline mean accuracy was 61% against ZeroGPT's 46.1%, yet Turnitin missed 88 of 104 AI samples and ZeroGPT 77. Neither tool falsely accused a human text in Weber-Wulff 2023 or Perkins 2024. ZeroGPT is free for 15,000 characters per detection with paid plans from C$22.99 per month; Turnitin is licensed to institutions only (zerogpt.com/pricing, guides.turnitin.com). All read 8 November 2026. No detector score should be the only evidence in a misconduct or hiring decision.
How do ZeroGPT and Turnitin compare on the 14 fields?
| Field | ZeroGPT | Turnitin |
|---|---|---|
| 1. Name, vendor, official domain | ZeroGPT, Olive Works LLC doing business as ZeroGPT, zerogpt.com | Turnitin AI writing detection, Turnitin, LLC, turnitin.com |
| 2. Modalities detected | text, image, video | text |
| 3. Generators covered | Chat GPT, GPT 3, GPT 4, GPT 5, GPT 6, Gemini, Grok, Perplexity AI, Claude, DeepSeek, LLaMa models, 35+ LLM families | 36 named models including Amazon Nova 2 Lite, Claude-3-Haiku through Claude Opus 4.6, DeepSeek V3.2, Gemini-1.0-pro through Gemini-3.1-pro-preview, GPT-4o through GPT-5.4-pro, o1-mini, Grok-4.1, Llama-3.3-70b, Llama-4-Maverick-17b, Mistral-large-3; Spanish model: GPT-3.5 and GPT-4; Japanese model: GPT-4 (0613), GPT-4o, GPT-4o-mini |
| 4. Languages supported | 7: English, Spanish, Japanese, Modern Arabic, Dutch, German, Turkish | |
| 5. Input limits | Maximum characters per AI detection: FREE 15,000, PRO 100,000, PLUS 100,000, MAX 150,000, EDU 250,000, EXPERT 350,000, ENTERPRISE unlimited. Batch files: FREE 0, PRO 50, PLUS 60, MAX 75, EDU 80, EXPERT 100. Paste text or upload .txt, .docx, text-based .pdf. Guidance: at least 150 to 200 words. API plans: 2MB (Beginner), 5MB (PRO), 15MB (VIP) max file size. | English: file size under 100 MB; at least 300 words of prose text in a long-form writing format; must not exceed 30,000 words; accepted file types .docx, .pdf, .txt, .rtf. Lists, headers, references, tables, formulas and code are excluded from scoring. |
| 6. Free tier | $0 forever, no credit card: 15,000 characters per AI detection, words allowance 1,250,000 /month, 0 batch files, no PDF reports, no detection history; plagiarism checker 300 words one-time-only | None. AI writing detection is only available to customers that license Turnitin Originality (or the AI writing add-on for iThenticate 2.0). |
| 7. Paid plans | PRO C$22.99 /month or C$13.99 /month billed annually; PLUS C$32.99 or C$20.99; MAX C$46.99 or C$29.99; EDU C$32.99 /month billed yearly only; EXPERT C$90.99 or C$57.99; ENTERPRISE get a quote; API pay as you go from $0.042 per 1,000 words | |
| 8. API | Yes, app.theneo.io ZeroGPT business API docs | |
| 9. Integrations | LMS or enterprise systems via API (no named LMS), WhatsApp bot, Telegram bot | LMS (Moodle, Blackboard, Canvas, etc.) via the Similarity Report; no AI indicator embedded in the LMS user interface; Microsoft Teams (Feedback Studio integration only) |
| 10. Vendor accuracy and false positive claims | States a detection accuracy rate of 98.4% and 35+ LLM families covered; under 1% false positive rate (product metrics); home FAQ: converge to an error rate lower than 2% | States a false positive rate under 1% for documents with over 20% AI writing, validated on over 700,000 pre-ChatGPT academic papers; ELL writers 0.014 and native English writers 0.013 in its ELL research; no accuracy rate stated |
| 11. Independent studies | 5 records: Weber-Wulff 2023, Liang 2023, Walters 2023, Perkins 2024, RAID 2024 | 3 independent records (Weber-Wulff 2023, Walters 2023, Perkins 2024) plus the vendor statements record |
| 12. Data retention and privacy statement | Privacy policy (October 07, 2026): no use of private user content to train or develop AI models; content retained only for the period reasonably necessary to provide and operate the service; infrastructure hosted in Germany under GDPR; no-retention modes for API and enterprise; paid plans include detection history | No separate repository for AI writing detection; customers choose whether student papers enter the repository; AI results retained as part of the similarity report; full deletion of submissions on request, no partial deletion of the AI component |
| 13. Launch year and ownership | Olive Works LLC, doing business as ZeroGPT, organized in Wyoming, United States; parent company not published | 2023; Turnitin, LLC; no parent company stated on pages read |
| 14. Rankings the card appears in | Independently tested, false positive rate, free tiers; listed in the unlimited table of input limits | Independently tested, false positive rate, free tiers, languages supported, input limits |
Which studies tested both ZeroGPT and Turnitin?
Three of the nine studies indexed on the morning of 9 November 2026 tested both tools, all in 2023 and 2024, and each is read in the primary paper (studies index, compiled 8 November 2026). Of the eight records added later that day, Van Vlasselaer 2026 tested Turnitin alone of the pair, scoring all 40 fully AI-generated papers under 20% (DOI 10.1007/s40979-026-00226-w, read 9 November 2026), and none tested ZeroGPT, so the shared count stays at three. ZeroGPT also appears in Liang 2023 and RAID 2024, which did not include Turnitin; Turnitin's own statements are kept in a separate vendor record.
| Study | ZeroGPT | Turnitin |
|---|---|---|
| Weber-Wulff 2023, 54 documents, tested March and April 2023 | 32 of 54 correct (59% binary accuracy, rank 6 of 14); 72% logarithmic accuracy (rank 5); 0 false accusations in 18 human-written documents; 13 of 36 AI documents undetected (36.1%) | 41 of 54 correct (76%, rank 1 of 14); 81% logarithmic accuracy; 0 false accusations in 18 human-written documents; the only tool to classify all unmodified ChatGPT documents correctly; 10 of 36 AI documents undetected (27.8%); Turnitin knew the documents were a test suite |
| Walters 2023, 126 documents, tested June and July 2023 | 87% of 126 correct, 1% incorrect, 12% uncertain; 100% on 42 GPT-3.5 papers, 83% on 42 GPT-4 papers; 79% of 42 human papers correct, 2% incorrect (false positive rate 2%) | 100% of 126 correct, 0 incorrect, 0 uncertain; 0 false positives in 42 human papers; 100% on both GPT-3.5 and GPT-4 papers; the author notes the papers could have been used to train the detector |
| Perkins 2024, 114 samples | Baseline mean accuracy 46.1% (rank 5 of 7); 31.3% on non-manipulated AI text falling to 17.3% after adversarial techniques (drop 14%); 0 false accusations in 10 human samples; 77 of 104 AI samples undetected (73%) | Baseline mean accuracy 61% (rank 2 of 7); 50% on non-manipulated AI text falling to 7.9% after adversarial techniques (drop 42.1%, the largest); 0 false accusations in 10 human samples; 88 of 104 AI samples undetected (84%) |
| Vendor statement, Turnitin claims record | Document false positive rate under 1% for documents with over 20% AI writing on over 700,000 pre-ChatGPT papers; sentence-level about 4% (2023); scores of 1% to 19% shown as an asterisk |
Results as each study reports them: arXiv 2306.15666, De Gruyter 10.1515/opis-2022-0158, arXiv 2403.19148, Turnitin FAQ, all read 8 November 2026.
The three studies point the same way on two fields and split on a third. On false accusations the two tools are level in Weber-Wulff 2023 and Perkins 2024 at 0, and Walters 2023 separates them by one human paper out of 42. On unedited AI text Turnitin led in all three. On edited text the order flips in Perkins 2024: Turnitin fell 42.1 points to 7.9% and ZeroGPT fell 14 points to 17.3%, so after adversarial techniques ZeroGPT detected more of the manipulated samples (arXiv 2403.19148, read 8 November 2026).
What do the two cost over 12 months?
| Plan | Monthly as stated | Annual as stated | Inclusions as stated | Checked on |
|---|---|---|---|---|
| ZeroGPT FREE | $0 forever | 15,000 characters per detection, 1,250,000 words per month, 0 batch files, no reports or history | 8 November 2026 | |
| ZeroGPT PRO | C$22.99 /month | C$13.99 /month billed annually = C$167.88 /year | 100,000 characters per detection, 50 batch files, PDF reports, history, 4,500,000 words/month | 8 November 2026 |
| ZeroGPT PLUS | C$32.99 /month | C$20.99 /month billed annually = C$251.88 /year | 100,000 characters, 60 batch files, 5,000,000 words/month, plagiarism 35,000 words/month | 8 November 2026 |
| ZeroGPT MAX | C$46.99 /month | C$29.99 /month billed annually = C$359.88 /year | 150,000 characters, 75 batch files, 7,500,000 words/month, plagiarism 60,000 words/month | 8 November 2026 |
| ZeroGPT EDU | C$32.99 /month, billed yearly only, C$395.88 /year | 250,000 characters, 80 batch files, 70,000,000 words/year, plagiarism 1,200,000 words/year | 8 November 2026 | |
| ZeroGPT EXPERT | C$90.99 /month | C$57.99 /month billed annually = C$695.88 /year | 350,000 characters, 100 batch files, 15,000,000 words/month, plagiarism 150,000 words/month | 8 November 2026 |
| ZeroGPT ENTERPRISE | Get a quote; unlimited characters per detection | 8 November 2026 | ||
| Turnitin AI writing detection | Included with a Turnitin Originality licence or the AI writing add-on for iThenticate 2.0; no price published; institution-level contract | 8 November 2026 |
ZeroGPT prices in CAD as shown to a Canadian viewer on zerogpt.com/pricing, checked 8 November 2026; a reader in another country may see different figures. Turnitin publishes no price on the FAQ or product page read the same day.
One of three comparisons in the compare hub. The cards are ZeroGPT and Turnitin, and the home board answers which AI detectors exist and who has tested them?
Which card fits which reader?
- Choose ZeroGPT ifyou have no institutional licence: the FREE plan takes 15,000 characters per detection and PRO costs C$22.99 per month billed monthly, while Turnitin sells no individual access (zerogpt.com/pricing, guides.turnitin.com, read 8 November 2026).
- Choose ZeroGPT ifyour texts are outside Turnitin's seven languages: the ZeroGPT FAQ states that all languages are supported, with no count published (zerogpt.com/faq, read 8 November 2026).
- Choose Turnitin ifyou grade inside Moodle, Blackboard or Canvas and your institution holds a Turnitin Originality licence: the AI indicator arrives in the Similarity Report, and three indexed studies recorded 0 false accusations (guides.turnitin.com, read 8 November 2026).
- Choose Turnitin ifunedited ChatGPT text is what you expect to see: 41 of 54 in Weber-Wulff 2023 and 126 of 126 in Walters 2023 against ZeroGPT's 32 of 54 and 87% (arXiv 2306.15666, De Gruyter).
- IfRead Perkins 2024 before relying on either for paraphrased text: Turnitin detected 7.9% and ZeroGPT 17.3% of AI text after adversarial techniques (arXiv 2403.19148, read 8 November 2026).
How does the evidence compare, tool by tool?
ZeroGPT
- Register ID
- AIDI-D-0011
- Study records
- 7 in the index: Weber-Wulff 2023, Liang 2023, Walters 2023, Perkins 2024, RAID 2024, Pegoraro 2023, Kar 2024
- Shared with Turnitin
- 3: Weber-Wulff 2023, Walters 2023, Perkins 2024
- Vendor accuracy claim
- 98.4% Detection accuracy; 35+ LLM families covered (ZeroGPT product metrics, DeepAnalyse page and home). FAQ: pushing toward >98% on internal evaluations.
- Free tier
- $0 forever, no credit card: 15,000 characters per AI detection, words allowance 1,250,000 /month, 0 batch files, no PDF reports, no detection history; plagiarism checker 300 words one-time-only
- Languages
- Not stated
Turnitin
- Register ID
- AIDI-D-0002
- Study records
- 5 in the index: Weber-Wulff 2023, Walters 2023, Perkins 2024, Van Vlasselaer 2026, Turnitin claims
- Shared with ZeroGPT
- 3: Weber-Wulff 2023, Walters 2023, Perkins 2024
- Vendor accuracy claim
- None published
- Free tier
- None. AI writing detection is only available to customers that license Turnitin Originality (or the AI writing add-on for iThenticate 2.0).
- Languages
- 7 stated
How this comparison is built
- Fields
- The 14 fields of every card, in card order
- Shared study records
- 3: Weber-Wulff 2023, Walters 2023, Perkins 2024
- Fields checked
- 8 November 2026
- Records in the studies index
- 35 records
Both columns are drawn from the same 14 fields in the index data file, read on 8 November 2026 from the vendor page recorded on each card, and from the studies index, where each result is copied as the paper reports the figure, with the base stated. The three shared studies are the only rows where the two tools can be compared on one sample; a vendor's own statement is labelled as such and never set beside a study figure as its equal. The rules, the recheck schedule (next pass 8 December 2026) and the correction route are on the methodology page; the independently tested ranking sorts every card on study count.
Two later comparisons read the same records for Turnitin against other tools: Winston AI vs Turnitin on Weber-Wulff 2023 and Scribbr vs Turnitin on Walters 2023, both added 9 November 2026.
Where each figure comes from
Values as each vendor states them on zerogpt.com, zerogpt.com/pricing, zerogpt.com/deep-analyse, zerogpt.com/privacy-policy, zerogpt.com/faq, the Turnitin AI writing FAQ and turnitin.com, all read 8 November 2026. ZeroGPT's prices were shown in CAD to a Canadian viewer with no currency switcher. An empty cell means the card has no published value for the field.
Questions about ZeroGPT and Turnitin
How many studies tested both ZeroGPT and Turnitin?
Three: Weber-Wulff 2023 (54 documents), Walters 2023 (126 documents) and Perkins 2024 (114 samples), all read in the primary paper and compiled 8 November 2026 (studies index). ZeroGPT also appears in Liang 2023 and RAID 2024.
Which had fewer false positives?
Both recorded 0 false accusations in 18 human-written documents in Weber-Wulff 2023 and 0 in 10 in Perkins 2024; in Walters 2023 Turnitin flagged 0 of 42 human papers and ZeroGPT 2% (arXiv 2306.15666, arXiv 2403.19148, De Gruyter, read 8 November 2026).
Which caught more AI text?
Turnitin on unedited text: 126 of 126 in Walters 2023 against ZeroGPT's 87%, and 26 of 36 AI documents in Weber-Wulff 2023 against ZeroGPT's 23 of 36. ZeroGPT after edits: 17.3% against Turnitin's 7.9% in Perkins 2024 (De Gruyter, arXiv 2306.15666, arXiv 2403.19148, read 8 November 2026).
Is ZeroGPT free?
Yes, with limits: the FREE plan states 15,000 characters per detection and 1,250,000 words per month, with no batch files, reports or history (zerogpt.com/pricing, read 8 November 2026). Turnitin has no free tier.
Can a student use Turnitin's AI detection?
Only through an institution: the FAQ states that AI writing detection is available to customers licensing Turnitin Originality or the iThenticate 2.0 add-on, read 8 November 2026 (guides.turnitin.com).
Why are ZeroGPT's prices in Canadian dollars?
The pricing page geo-priced to a Canadian viewer on 8 November 2026 and offered no currency switcher, so the card records the figures as shown, in CAD (zerogpt.com/pricing). The API pay-as-you-go rates on the same page are shown in USD.
What minimum text does each need?
ZeroGPT guides users to at least 150 to 200 words; Turnitin requires 300 words of prose and shows scores of 1% to 19% as an asterisk, both read 8 November 2026 (zerogpt.com/pricing, guides.turnitin.com).
Did Turnitin know it was being tested?
In Weber-Wulff 2023, yes: the paper states that Turnitin knew the documents were part of a test suite, and the test ran April 14 to 20, 2023, after the other tools (arXiv 2306.15666, read 8 November 2026). Walters 2023 notes the student papers could have been used to train the Turnitin detector.
Sources and revisions
Sources
- ZeroGPTzerogpt.com
- ZeroGPT pricingzerogpt.com/pricing
- ZeroGPT DeepAnalyse pagezerogpt.com/deep-analyse
- ZeroGPT FAQzerogpt.com/faq
- ZeroGPT privacy policyzerogpt.com/privacy-policy
- Turnitin AI writing FAQguides.turnitin.com/hc/en-us/articles/284...
- Turnitin AI writing product pageturnitin.com/solutions/topics/ai-writing
- Weber-Wulff et al. 2023arxiv.org/pdf/2306.15666
- Walters 2023degruyterbrill.com/document/doi/10.1515/o...
- Perkins et al. 2024arxiv.org/pdf/2403.19148
Revision log
- Comparison created
- Winston ai vs turnitin and scribbr vs turnitin comparisons linked (fifth wave)
