ZeroGPT vs Scribbr vs QuillBot: three free AI detectors on the same 14 fields
ZeroGPT carries 6 study records in the index, Scribbr 1 and QuillBot 0, and the one record that tested two of the three on the same documents is Walters 2023, where ZeroGPT classified 87% of 126 documents correctly and Scribbr 88%, each with a 2% false positive rate on 42 human-written papers (De Gruyter 10.1515/opis-2022-0158, Table 4, read 8 November 2026). All three offer a free check without an account, and the three free tiers are stated in three different units: 15,000 characters per detection and 1,250,000 words a month for ZeroGPT, "unlimited" for Scribbr, and 1,200 words per scan with 6 scans a day for QuillBot.
Updated Fields checked 8 November 2026
Short answer
ZeroGPT is the only one of the three with an API and more than one study record: six records, including 0 false accusations in 18 human documents in Weber-Wulff 2023 and a 45.7% true positive rate, the highest of 12 online tools, in Pegoraro 2023 (arXiv 2306.15666, arXiv 2304.01487). Scribbr's one record is Walters 2023, 88% of 126 documents correct with 2% of 42 human papers flagged; QuillBot has none, and its "99% detection rate, according to independent evaluations from RAID" is the vendor's statement (quillbot.com, read 8 November 2026). Scribbr states "we do not store or share your data"; ZeroGPT states content is retained only as long as needed to operate the service; QuillBot's detector page states no retention clause. Paid plans: ZeroGPT PRO C$22.99 a month, Scribbr $19.95 per premium check, QuillBot Premium $8.33 a month billed annually (zerogpt.com/pricing, scribbr.com, quillbot.com/premium, read 8 November 2026). No detector score should be the only evidence in a misconduct or hiring decision.
How do ZeroGPT, Scribbr and QuillBot compare on the 14 fields?
| Field | ZeroGPT | Scribbr | QuillBot |
|---|---|---|---|
| 1. Name, vendor, official domain | ZeroGPT, Olive Works LLC doing business as ZeroGPT, zerogpt.com | Scribbr's AI Detector, Scribbr, scribbr.com | AI Detector by Quillbot, QuillBot, quillbot.com |
| 2. Modalities detected | text, image, video | text | text, image |
| 3. Generators covered | Chat GPT, GPT 3, GPT 4, GPT 5, GPT 6, Gemini, Grok, Perplexity AI, Claude, DeepSeek, LLaMa models, "35+ LLM families" | ChatGPT, Gemini, Copilot, GPT2, GPT3, GPT3.5, GPT4 (experimental in the free tool, supported in Premium) | GPT-6, GPT-5, ChatGPT, Gemini, Claude, Llama, other LLMs |
| 4. Languages supported | "all languages", no count published | 1: English | 6: English, French, Spanish, German, Dutch, Portuguese |
| 5. Input limits | 15,000 characters per detection free, 100,000 on PRO and PLUS, 150,000 MAX, 250,000 EDU, 350,000 EXPERT, unlimited Enterprise; paste or upload .txt, .docx, text-based .pdf; guidance of at least 150 to 200 words | Minimum 80 words per scan; free up to 1,200 words per scan; Premium no word limit; paste or upload a document (file types not published); Premium bulk upload | |
| 6. Free tier | $0 forever, no credit card: 15,000 characters per detection, 1,250,000 words a month, no batch files, no PDF reports, no history | Unlimited free AI checks, no sign-up required | Up to 1,200 words per scan, up to 6 scans a day, 1 free explainer card, no sign-up under 1,200 words |
| 7. Paid plans | PRO C$22.99 a month or C$13.99 billed annually; PLUS C$32.99 or C$20.99; MAX C$46.99 or C$29.99; EDU C$32.99 billed yearly only; EXPERT C$90.99 or C$57.99; Enterprise by quote; API from $0.042 per 1,000 words | No detector subscription: premium plagiarism check with the premium AI Detector at $19.95 (up to 7,499 words), $29.95 (7,500 to 49,999 words) or $39.95 (50,000+ words) per check | Premium $8.33 USD a month billed annually, detector bundled with the writing tools, no word limit and unlimited scans; Team Plan price not shown |
| 8. API | Yes, app.theneo.io (ZeroGPT Business API) | ||
| 9. Integrations | LMS or enterprise systems via API (no named LMS), WhatsApp bot, Telegram bot | Google Chrome extension, Microsoft Edge extension, macOS app, Word add-in, Android and iOS keyboard and app | |
| 10. Vendor accuracy and false positive claims | "98.4% Detection accuracy", "35+ LLM families covered"; "<1% False positive rate"; home FAQ: "converge to an error rate lower than 2%" | "Unfortunately, we can't guarantee 100% accuracy"; the vendor's own research found "the highest accuracy we found was 84% in a premium tool or 68% in the best free tool" | "Detect AI text reliably with a 99% detection rate, according to independent evaluations from RAID"; "when the result is unclear, our model tends to classify texts as human written, which reduces false positives", no numeric rate |
| 11. Independent studies | 6 records: Weber-Wulff 2023, Liang 2023, Walters 2023, Perkins 2024, RAID 2024, Pegoraro 2023 | 1 record: Walters 2023 | 0 records |
| 12. Data retention and privacy statement | Privacy policy (7 October 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 in Germany under GDPR; no-retention modes for API and enterprise; paid plans keep a detection history | "we do not store or share your data. Submissions will never be added to our content database" | |
| 13. Launch year and ownership | Not stated; Olive Works LLC, organized in Wyoming, parent not published | Not stated; part of QuillBot, within the Learneo family of businesses | Not stated; part of the Learneo family of businesses (stated on scribbr.com) |
| 14. Rankings the card appears in | Independently tested, false positive rate, free tiers, adversarial robustness, false negative rate, claim vs evidence gap, studies by content type, platform integrations, file types; unlimited table of input limits | Independently tested, false positive rate, languages, data retention | Languages, studies by content type, platform integrations |
Which studies tested two or more of the three?
One record. Walters 2023 submitted 126 documents (42 human-written first-year composition papers, 42 GPT-3.5 papers and 42 GPT-4 papers) to 16 detectors between 25 June and 12 July 2023 and printed the share correct, incorrect and uncertain per tool (De Gruyter 10.1515/opis-2022-0158, Table 4, read 8 November 2026). ZeroGPT and Scribbr were both in that set; QuillBot was not, and no other indexed record tested it. ZeroGPT's other five records tested it alone among the three: Weber-Wulff 2023, Liang 2023, Perkins 2024, RAID 2024 and Pegoraro 2023.
| Source | ZeroGPT | Scribbr | QuillBot |
|---|---|---|---|
| Walters 2023, share of 126 documents correct (Table 4) | 87% correct, 1% incorrect, 12% uncertain | 88% correct, 11% incorrect, 1% uncertain | Not tested |
| Walters 2023, 42 GPT-3.5 and 42 GPT-4 papers | 100% on GPT-3.5, 83% on GPT-4 | 100% on GPT-3.5, 69% on GPT-4 | Not tested |
| Walters 2023, 42 human-written papers | 79% correct, 2% incorrect (false positive rate 2%), the rest uncertain | 95% correct, 2% incorrect (false positive rate 2%) | Not tested |
| Weber-Wulff 2023, 54 documents | 32 of 54 correct (59%, rank 6 of 14); 0 false accusations in 18 human documents; 13 of 36 AI documents undetected | Not tested | Not tested |
| Liang 2023, 91 TOEFL and 88 US essays | One of seven detectors averaging a 61.22% false positive rate on TOEFL essays; per-detector rates in a chart only | Not tested | Not tested |
| Perkins 2024, 114 samples | Baseline mean accuracy 46.1% (rank 5 of 7); 31.3% on unmodified AI text falling to 17.3% after adversarial techniques; 0 false accusations in 10; 77 of 104 AI samples undetected | Not tested | Not tested |
| RAID 2024, shared benchmark | False positive rate 1.71%, 1.42%, 1.21% and 0.90% at thresholds 0.25 to 0.95; 95.4% accuracy at 5% FPR on open-source chat models, 0.3% on non-chat with sampling and a repetition penalty | Not tested | Not tested |
| Pegoraro 2023, ChatGPT and human Q&A benchmark (Table I) | True positive rate 45.7%, the highest of the 12 online tools; true negative rate 92.2% | Not tested | Not tested |
| Vendor statement, quillbot.com (read 8 November 2026) | "99% detection rate, according to independent evaluations from RAID"; no false positive rate |
Results as each study reports them: De Gruyter 10.1515/opis-2022-0158, arXiv 2306.15666, arXiv 2304.02819, arXiv 2403.19148, arXiv 2405.07940, arXiv 2304.01487, read 8 and 9 November 2026. "Not tested" means the paper did not run that tool.
One of the comparisons in the compare hub. The cards are ZeroGPT, Scribbr and QuillBot, and the home board answers which AI detectors exist and who has tested them? Scribbr and QuillBot share an owner, Learneo, which is why their two columns read alike on several fields.
Which card fits which reader?
- Choose ZeroGPT ifyou want the free check with the largest stated allowance: 15,000 characters per detection and 1,250,000 words a month, against QuillBot's 1,200 words per scan (zerogpt.com/pricing, quillbot.com, read 8 November 2026).
- Choose ZeroGPT ifyou need an API: it is the only one of the three with one, metered from $0.042 per 1,000 words (zerogpt.com/pricing, read 8 November 2026).
- Choose Scribbr ifa written no-storage line matters: "we do not store or share your data. Submissions will never be added to our content database" (scribbr.com/ai-detector, read 8 November 2026); ZeroGPT states retention for as long as needed to operate, and QuillBot's page states nothing.
- Choose Scribbr ifyou want the vendor that publishes its own ceiling: its page says no tool reached more than 84% accuracy in its research and declines to guarantee any figure (scribbr.com/ai-detector, read 8 November 2026).
- Choose QuillBot ifyou check French, Spanish, German, Dutch or Portuguese text and want the detector inside a Word add-in or a phone keyboard: six languages and six integrations are stated (quillbot.com/ai-content-detector, quillbot.com/premium, read 8 November 2026); Scribbr states English only.
- Choose ZeroGPT ifyou want the only one of the three with results from more than one study: six records, from 0 of 18 false accusations in Weber-Wulff 2023 to 1.42% at threshold 0.5 in RAID 2024 (arXiv 2306.15666, arXiv 2405.07940).
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 Scribbr and QuillBot
- 0
- 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
Scribbr
- Register ID
- AIDI-D-0009
- Study records
- 1 in the index: Walters 2023
- Shared with ZeroGPT and QuillBot
- 0
- Vendor accuracy claim
- Unfortunately, we can't guarantee 100% accuracy. Our research into the best AI detectors indicates that no tool can provide complete accuracy; the highest accuracy we found was 84% in a premium tool or 68% in the best free tool. Free detector: GPT2, GPT3, and GPT3.5 with average accuracy.
- Free tier
- Unlimited free AI checks, no sign-up required
- Languages
- 1 stated
QuillBot
- Register ID
- AIDI-D-0007
- Study records
- 2 in the index: Kar 2024, Alshammari and Rao 2025
- Shared with ZeroGPT and Scribbr
- 0
- Vendor accuracy claim
- Detect AI text reliably with a 99% detection rate, according to independent evaluations from RAID.
- Free tier
- Up to 1,200 words per scan, up to 6 scans per day, 1 free explainer card; no sign-up for scans under 1,200 words
- Languages
- 6 stated
How this comparison is built
- Fields
- The 14 fields of every card, in card order
- Shared study records
- 0
- Fields checked
- 8 November 2026
- Records in the studies index
- 35 records
All three 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. Walters 2023 is the one record that ran two of the three on the same documents, so its rows sit first in the evidence table; the single-tool records follow with "Not tested" in the other columns, and QuillBot's vendor statement sits in a labelled row of its own. The rules, the recheck schedule (next pass 8 December 2026) and the correction route are on the methodology page; the free tiers ranking explains why only a stated monthly word count is ranked and why Scribbr's "unlimited" and QuillBot's per-day limit sit in its second table.
Where each figure comes from
Values as each vendor states them on zerogpt.com, zerogpt.com/pricing, zerogpt.com/faq, zerogpt.com/privacy-policy, scribbr.com/ai-detector, scribbr.com/about-us, quillbot.com/ai-content-detector and quillbot.com/premium, all read 8 November 2026. An empty cell means the card has no published value for the field. ZeroGPT's prices were geo-priced in Canadian dollars to the reader on that date with no currency switcher.
What do the three cost over 12 months?
Prices as shown on zerogpt.com/pricing (Canadian dollars to the reader, no USD shown except the API table), scribbr.com/ai-detector and scribbr.com/plagiarism-checker (USD), and quillbot.com/premium (USD), checked 8 November 2026. Scribbr sells per check and QuillBot bundles the detector into a writing subscription, so the three columns are not like for like.
Questions about ZeroGPT, Scribbr and QuillBot
Which of the three has been tested by a study?
ZeroGPT in six records and Scribbr in one; QuillBot in none, as compiled on 9 November 2026 (studies index). The one shared record is Walters 2023, where Scribbr scored 88% of 126 documents and ZeroGPT 87%.
Which free detector has the largest limit?
ZeroGPT states 15,000 characters per detection and 1,250,000 words a month; Scribbr states unlimited free checks with no number; QuillBot states 1,200 words per scan and 6 scans a day (zerogpt.com/pricing, scribbr.com, quillbot.com, read 8 November 2026).
Which has the lower false positive rate?
Walters 2023 recorded 2% of 42 human papers for both ZeroGPT and Scribbr; ZeroGPT also recorded 0 of 18 in Weber-Wulff 2023 and 1.42% at threshold 0.5 in RAID 2024 (De Gruyter, arXiv 2306.15666, arXiv 2405.07940). QuillBot publishes no rate and no study measured one.
Are Scribbr and QuillBot the same detector?
Both are Learneo businesses and Scribbr describes itself as part of QuillBot (scribbr.com/about-us, read 8 November 2026). The two pages state different language lists, limits and claims, so the index keeps two cards; the pairing is recorded on the shared owner page.
Which supports more languages?
QuillBot names 6, Scribbr 1 and ZeroGPT says all languages without a count (quillbot.com, scribbr.com, zerogpt.com/faq, read 8 November 2026). The languages ranking sorts only stated counts, so ZeroGPT is in its second table.
Which keeps the text you paste?
Scribbr states it does not store submissions; ZeroGPT states content is kept only for the period reasonably necessary to operate the service, with detection history on paid plans; QuillBot's detector page states no retention clause (scribbr.com, zerogpt.com/privacy-policy, read 8 November 2026).
What does each cost beyond the free tier?
ZeroGPT PRO C$22.99 a month or C$13.99 billed annually; Scribbr $19.95 to $39.95 per premium check with no subscription; QuillBot Premium $8.33 USD a month billed annually (zerogpt.com/pricing, scribbr.com, quillbot.com/premium, checked 8 November 2026).
Sources and revisions
Sources
- ZeroGPTzerogpt.com
- ZeroGPT pricingzerogpt.com/pricing
- ZeroGPT FAQzerogpt.com/faq
- ZeroGPT privacy policyzerogpt.com/privacy-policy
- Scribbr AI Detectorscribbr.com/ai-detector
- Scribbr plagiarism checker (premium prices)scribbr.com/plagiarism-checker
- Scribbr aboutscribbr.com/about-us
- QuillBot AI Detectorquillbot.com/ai-content-detector
- QuillBot Premiumquillbot.com/premium
- Walters 2023degruyterbrill.com/document/doi/10.1515/o...
- Weber-Wulff et al. 2023arxiv.org/pdf/2306.15666
- Liang et al. 2023arxiv.org/pdf/2304.02819v3
- Perkins et al. 2024arxiv.org/pdf/2403.19148
- Dugan et al. 2024 (RAID)arxiv.org/pdf/2405.07940
- Pegoraro et al. 2023 (read 9 November 2026)arxiv.org/abs/2304.01487
