Pangram vs Originality.ai: the 14 fields and the study results side by side
Originality.ai carries 6 study records in the index and Pangram 2, and one record ran both on the same texts: Jabarian and Imas 2025, where Pangram recorded a false positive rate of 0.0000 and Originality.ai 0.0011 on 1,992 human passages at a 0.5 threshold (NBER w34223, Table 3, read 9 November 2026). The two cards are read side by side below on the same 14 fields, with the vendor's wording and the page it came from, so the comparison rests on what each company publishes and what outside researchers measured.
Updated Fields checked 8 November 2026
Short answer
In the one shared study, Jabarian and Imas 2025, Pangram's false positive rate was 0.0000 and Originality.ai's 0.0011 on 1,992 human passages at threshold 0.5, and after one StealthGPT rewrite Pangram's false negative rate ran 0.0000 to 0.0500 against Originality.ai's 0.0227 to 0.2073 (NBER w34223, Tables 3 and B.4). Originality.ai brings four more records, including 0 of 42 human papers flagged in Walters 2023 and 0.25% false positives at threshold 0.5 in RAID 2024; Pangram brings Van Vlasselaer 2026, where it placed 37 of 40 humanised papers within their AI range (De Gruyter 10.1515/opis-2022-0158, arXiv 2405.07940, DOI 10.1007/s40979-026-00226-w). Pangram states 24 named languages, text, image and video; Originality.ai states 30 languages without a list, text and code. Pangram Individual is $20 a month; Originality.ai Pro is $14.95 a month or $12.95 billed yearly (pangram.com/pricing, originality.ai/pricing, read 8 November 2026). No detector score should be the only evidence in a misconduct or hiring decision.
How do Pangram and Originality.ai compare on the 14 fields?
| Field | Pangram | Originality.ai |
|---|---|---|
| 1. Name, vendor, official domain | Pangram, Pangram Labs, pangram.com | AI Detector By Originality.ai, Originality AI Inc., originality.ai |
| 2. Modalities detected | text, image, video | text, code |
| 3. Generators covered | OpenAI, Anthropic, Google, Meta, DeepSeek, xAI; Claude Opus 5, GPT 5.6, Gemini 3, Grok 4.3, Claude Sonnet 5, Claude Fable 5, GPT-5.4, GPT-OSS 120B, DeepSeek V4, Llama 3.3, Mistral Medium 3.5, Kimi K2.6, Qwen 3.7 Max, Gemma 4, Nemotron 3 Ultra, GLM 5.2; 26 models benchmarked | GPT-6 Astra (Sol, Luna, Terra), GPT-6.1 Sol, Claude Fable 5, Claude Opus 5.5, Gemini 3, Gemini 4 Argon, Kimi K3, DeepSeek V4 (Flash and Pro), Grok 4.7, Llama 4 Maverick and Scout, "and more" |
| 4. Languages supported | 24 named: English, Arabic, Chinese, Czech, Dutch, French, German, Greek, Hindi, Hungarian, Italian, Japanese, Korean, Persian, Polish, Portuguese, Romanian, Russian, Spanish, Swedish, Turkish, Ukrainian, Urdu, Vietnamese | 30 stated ("30+ languages", internal test across 30 languages); no list published |
| 5. Input limits | PDF, DOCX, RTF files and up to 100 files at a time; minimum and maximum words per text check not published; image JPG, PNG, WebP, minimum 512x512, up to 30 MB; video MP4, MOV, WebM, minimum 720x720, up to 500 MB; API 5 QPS realtime, bulk 1,000 billable units | 100-word minimum; 1 credit = 100 words; paste, upload (docx, pdf, doc) or URL; full site scans; maximum words per scan and file size not published |
| 6. Free tier | 20 free credits daily after signing up for the free plan, including AI assistance detection | Free, no subscription, $0: 10K words (100 credits) one time |
| 7. Paid plans | Individual $20 a month (300,000 words a month, 100 image scans, MCP, Google Docs and Gmail, $10 in API credits; 7-day free trial); Professional $65; Team $40 for 2 seats; API text $0.05 per 100 words, image $0.25 or $0.05, video $0.05 per second | Pro $14.95 a month or $12.95 billed yearly (200K words a month, Bulk Scan, 30-day history); Enterprise $179 a month or $136.58 billed yearly (1.5M words, API access, 1-year history); credits 1 per 100 words, 5.0 Max 10 credits, Deep Scan 5 credits |
| 8. API | Yes, docs.pangram.com | Yes, docs.originality.ai (Enterprise plan) |
| 9. Integrations | Chrome extension, Firefox extension, Google Docs, Gmail, Google Classroom, Canvas, Moodle, Brightspace (D2L), MCP server | Chrome Extension (Google Docs, Writing Replay), Firefox Extension, Google Classroom via the Chrome extension, Moodle Plugin 2.0, API, MCP, Bulk Scan |
| 10. Vendor accuracy and false positive claims | "accuracy of over 99%" with third-party verification by researchers at the University of Chicago and The University of Maryland; "accurate on 99.6% of these AI-generated samples"; false positive rate 1 in 10,000 on an aggregate of public datasets of tens of millions of documents | "Originality.ai 5.0 and 5.0 Max have 99%+ accuracy" (blog, 25 October 2026); AI recall 97.69% (5.0) and 98.00% (5.0 Max) on UChicago Booth AI; human specificity 99.954% and 99.958% on PERSUADE 2.0; historic rates 1.5% (3.0.2 Turbo), 0.5% (1.0.2 Lite), 2.4% (Multilingual 2.0.0) |
| 11. Independent studies | 2 records: Jabarian and Imas 2025, Van Vlasselaer 2026 | 6 records: Liang 2023, Walters 2023, RAID 2024, Orenstrakh 2023, Jabarian and Imas 2025, Pegoraro 2023 |
| 12. Data retention and privacy statement | Submitted text stored in the dashboard and extension while the account is active, deletable from the history tab; detection trained on a proprietary dataset that excludes user submissions; never shared or sold; zero data retention for the enterprise API | Privacy policy (17 June 2026): scan history and results may train the models unless the user opts out; no personal or student data used for training; retained only as long as reasonably necessary; some features call OpenAI's API, which may keep text in abuse logs for up to 30 days |
| 13. Launch year and ownership | 2023; Pangram Labs, co-founders Max Spero and Bradley Emi, Brooklyn, NY; no parent company stated | 2022; Originality AI Inc., Collingwood, Ontario, founder Jon Gillham, co-founder and CTO Conor Watt; parent not published |
| 14. Rankings the card appears in | Independently tested, false positive rate, languages, adversarial robustness, false negative rate, claim vs evidence gap, studies by content type, platform integrations, file types | Independently tested, false positive rate, languages, adversarial robustness, false negative rate, claim vs evidence gap, studies by content type, platform integrations, file types |
Which studies tested both Pangram and Originality.ai?
One record. Jabarian and Imas 2025, an NBER working paper, submitted 1,992 human passages written before 2020 (Amazon reviews, blog posts, news, novels, restaurant reviews and resumes) and 7,968 AI passages from GPT-4.1, Claude Opus 4, Claude Sonnet 4 and Gemini 2.0 Flash to the commercial APIs of Pangram, Originality.ai and GPTZero and to a RoBERTa baseline, and printed every rate at fixed thresholds (NBER w34223, read 9 November 2026). Originality.ai's other five records tested it without Pangram: Liang 2023, Walters 2023, RAID 2024, Orenstrakh 2023 and Pegoraro 2023. Pangram's other record, Van Vlasselaer 2026, tested it without Originality.ai.
| Source | Pangram | Originality.ai |
|---|---|---|
| Jabarian and Imas 2025, false positive rate at threshold 0.5, 1,992 human passages (Table 3) | 0.0000 | 0.0011 |
| Jabarian and Imas 2025, false positive rate at the loosest threshold, 0.1 (Table 3) | 0.0010 | 0.0027 |
| Jabarian and Imas 2025, false positive rate at Youden thresholds, highest genre (Table 6) | 0.0075 | 0.0218 |
| Jabarian and Imas 2025, false negative rate at threshold 0.9, highest model (Table 4) | 0.0382 | 0.4242 (Claude Opus 4) |
| Jabarian and Imas 2025, false negative rate after one StealthGPT rewrite (Table B.4) | 0.0000 to 0.0500 | 0.0227 to 0.2073 |
| Jabarian and Imas 2025, false negative rate at a 0.0001 false positive cap (Table C.2) | 0.0095 to 0.0211 | 0.0749 to 0.5129 |
| Jabarian and Imas 2025, cost per correctly flagged AI passage, six-genre average (Table B.6) | $0.0228 | $0.0416 |
| Van Vlasselaer 2026, 160 papers | 65% strict and 97.5% inclusive accuracy on 40 fully AI papers; 37 of 40 hybrid and 37 of 40 humanised papers within range; 0 of 40 human papers flagged; 45.5% of 1,163 theses flagged when run alone | Not tested |
| Liang 2023, 91 TOEFL and 88 US essays | Not tested | One of seven detectors averaging a 61.22% false positive rate on TOEFL essays; per-detector rates in a chart only |
| Walters 2023, 126 documents | Not tested | 98% correct (model 1.4), 0 of 42 human papers flagged, 2 uncertain; 100% on both GPT-3.5 and GPT-4 papers |
| RAID 2024, shared benchmark | Not tested | False positive rate 0.47%, 0.25%, 0.17% and 0.07% at thresholds 0.25 to 0.95; 85.0% with no attack, 96.7% under paraphrase, 9.3% under the homoglyph attack at 5% FPR |
| Orenstrakh 2023, 164 submissions | Not tested | 93.86% on 114 human submissions, 7 false positives; 70.00% on 20 ChatGPT submissions; 100% to 40% on 10 after QuillBot |
| Pegoraro 2023, ChatGPT and human Q&A benchmark (Table I) | Not tested | True positive rate 7.6%, true negative rate 95.0%; the authors note the tool "is prone to classify ChatGPT-generated content as real" |
Results as each study reports them: NBER w34223, DOI 10.1007/s40979-026-00226-w, arXiv 2304.02819, De Gruyter 10.1515/opis-2022-0158, arXiv 2405.07940, arXiv 2307.07411, arXiv 2304.01487, read 8 and 9 November 2026. "Not tested" means the paper did not run that tool.
What do the two cost over 12 months?
| Plan | Monthly as stated | Annual as stated | Inclusions as stated | Checked on |
|---|---|---|---|---|
| Pangram Free | $0 | 20 free credits daily after signing up | 8 November 2026 | |
| Pangram Individual | $20 / month | Save $60, annually (12 x $20 less $60 is $180 by arithmetic) | 300,000 words a month, 100 image scans, similarity detection, MCP, Google Docs and Gmail, $10 in monthly API credits; 7-day free trial | 8 November 2026 |
| Pangram Professional | $65 / month | Save $240, annually ($540 by arithmetic) | 5x more words, 5x more image scans, $200 of API credits monthly | 8 November 2026 |
| Pangram Team | $40 / month, 2 seats | $20 / seat per month | 8 November 2026 | |
| Pangram API | Text (Pangram 4.1) $0.05 per 100 words; Image 1.0 $0.25 per image; Image Flash 1.0 $0.05 per image; Video $0.05 per second | Bulk text API: 20% discount | 8 November 2026 | |
| Originality.ai Free | $0 | 10K words (100 credits) one time | 8 November 2026 | |
| Originality.ai Pro | $14.95/month billed monthly | $12.95/month billed yearly, $155.40 a year | 200K words (2K credits) a month, Bulk Scan, 30-day scan history, team seats $8.62 or $9.95 a month | 8 November 2026 |
| Originality.ai Enterprise | $179/month billed monthly | $136.58/month billed yearly, $1,638.96 a year | 1.5M words (15K credits) a month, API access, 1-year scan history, team seats $19.04 or $24.95 a month | 8 November 2026 |
| Originality.ai credits | 1 credit per 100 words ($0.01 per 100 words, $0.10 per 1,000); 5.0 Max 10 credits; Deep Scan 5 credits | Pay-as-you-go credits expire 2 years after purchase; subscription credits expire monthly | 8 November 2026 |
Prices as shown on pangram.com/pricing and originality.ai/pricing, checked 8 November 2026; Originality.ai names USD, Pangram shows the $ symbol only. The plans include different word allowances (300,000 against 200,000 a month on the first paid tier), so the monthly figures are not like for like.
One of the comparisons in the compare hub. The cards are Pangram and Originality.ai, and the home board answers which AI detectors exist and who has tested them? The three-way page with GPTZero sets every Jabarian and Imas figure for three tools in one table.
Which card fits which reader?
- Choose Pangram ifyour texts are in Arabic, Chinese, Japanese, Korean or another of its 24 named languages and you want the list in writing: Originality.ai states 30 languages but publishes no list (pangram.com, originality.ai, read 8 November 2026).
- Choose Pangram ifhumanised or rewritten AI text is what you check: 0.0000 to 0.0500 false negatives after StealthGPT rewrites against Originality.ai's 0.0227 to 0.2073 in Jabarian and Imas 2025, and 37 of 40 humanised papers within range in Van Vlasselaer 2026 (NBER w34223, Table B.4; DOI 10.1007/s40979-026-00226-w, Table 5).
- Choose Originality.ai ifyou want the longer record: six studies from 2023 to 2025, including RAID 2024's shared benchmark and Walters 2023's 0 of 42 human papers flagged (arXiv 2405.07940, De Gruyter).
- Choose Originality.ai ifyou check code or whole websites: the card states code detection, URL input and full site scans, none of which Pangram states (originality.ai, read 8 November 2026).
- Choose Pangram ifyou check images or video in the same account: both are stated with file specifications and metered API prices (pangram.com, read 8 November 2026).
- Choose Originality.ai ifyour check sits under 300,000 words a month and price per word matters: Pro is $12.95 to $14.95 a month for 200,000 words against Pangram Individual at $20 for 300,000, and the per-passage cost in Jabarian and Imas 2025 was $0.0416 against $0.0228 (originality.ai/pricing, pangram.com/pricing, NBER w34223, Table B.6).
- Choose Pangram iftraining on your submissions is a concern: the card states detection is trained on a proprietary dataset that excludes user submissions, while Originality.ai's policy allows scan history to train models unless the user opts out (pangram.com/data-privacy, originality.ai/privacy-policy, read 8 November 2026).
How does the evidence compare, tool by tool?
Pangram
- Register ID
- AIDI-D-0006
- Study records
- 4 in the index: Jabarian and Imas 2025, Van Vlasselaer 2026, DAMAGE 2025, Epoch AI 2026
- Shared with Originality.ai
- 2: Jabarian and Imas 2025, Epoch AI 2026
- Vendor accuracy claim
- Pangram states accuracy of over 99% and third-party verification by researchers at the University of Chicago and The University of Maryland. We're accurate on 99.6% of these AI-generated samples.
- Free tier
- 20 free credits daily (twenty free checks a day after signing up for the free plan, including AI assistance detection)
- Languages
- 24 stated
Originality.ai
- Register ID
- AIDI-D-0004
- Study records
- 7 in the index: Liang 2023, Walters 2023, RAID 2024, Orenstrakh 2023, Jabarian and Imas 2025, Pegoraro 2023, Epoch AI 2026
- Shared with Pangram
- 2: Jabarian and Imas 2025, Epoch AI 2026
- Vendor accuracy claim
- Originality.ai 5.0 and 5.0 Max have 99%+ accuracy (accuracy blog, October 25, 2026). 5.0 launch post: AI recall UChicago Booth AI 97.69% (5.0) / 98.00% (5.0 Max), Epoch AI 99.33%; 5.0 Max identified 90.1% of mixed documents.
- Free tier
- Free, no subscription required, $0: 10K words (100 credits one time)
- Languages
- 30 stated
How this comparison is built
- Cards compared
- Pangram and Originality.ai
- Fields
- The 14 fields of every card, in card order
- Shared study records
- 2: Jabarian and Imas 2025, Epoch AI 2026
- 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. Jabarian and Imas 2025 is the one record that ran both tools on the same passages, so its rows sit first in the evidence table; the single-tool records follow with "Not tested" in the other column. A vendor's accuracy or false positive statement is quoted and attributed and never placed in the same cell as a study figure. The rules, the recheck schedule (next pass 8 December 2026) and the correction route are on the methodology page; the false negative rate ranking sorts both cards on the Jabarian and Imas and Walters figures used above.
Where each figure comes from
Values as each vendor states them on pangram.com, pangram.com/pricing, pangram.com/data-privacy, pangram.com/about-us, originality.ai, originality.ai/pricing, originality.ai/privacy-policy, originality.ai/extensions and the Originality.ai 5.0 post, all read 8 November 2026. An empty cell means the card has no published value for the field.
Questions about Pangram and Originality.ai
Which study tested both Pangram and Originality.ai?
Jabarian and Imas 2025, on 1,992 human and 7,968 AI passages through both APIs: false positive rate 0.0000 for Pangram and 0.0011 for Originality.ai at threshold 0.5 (NBER w34223, Table 3, read 9 November 2026).
Which has the lower false positive rate?
In the shared study Pangram, 0.0000 against 0.0011 of 1,992 passages. Originality.ai's other results are 0 of 42 in Walters 2023, 0.25% at threshold 0.5 in RAID 2024 and 7 of 114 in Orenstrakh 2023; Pangram's other result is 0 of 40 in Van Vlasselaer 2026 (De Gruyter, arXiv 2405.07940, arXiv 2307.07411, DOI 10.1007/s40979-026-00226-w).
Which catches rewritten AI text better?
In Jabarian and Imas 2025, after one StealthGPT rewrite Pangram missed 0.0000 to 0.0500 of AI passages and Originality.ai 0.0227 to 0.2073 (Table B.4); in RAID 2024 Originality.ai scored 96.7% under the paraphrase attack and 9.3% under the homoglyph attack, a test Pangram was not in (NBER w34223, arXiv 2405.07940).
Which supports more languages?
Originality.ai states 30 and Pangram names 24; only Pangram publishes the list (originality.ai, pangram.com, read 8 November 2026).
Which costs less per month?
Originality.ai Pro is $14.95 a month or $12.95 billed yearly for 200,000 words; Pangram Individual is $20 a month for 300,000 words (originality.ai/pricing, pangram.com/pricing, checked 8 November 2026). The allowances differ, so the prices are not like for like.
Do both have an API?
Yes: Pangram documents one at docs.pangram.com with metered prices from $0.05 per 100 words; Originality.ai's API comes with the Enterprise plan and is documented at docs.originality.ai (pangram.com/pricing, originality.ai/pricing, read 8 November 2026).
Which one detects code or images?
Originality.ai states code detection and Pangram states image and video detection; neither states the other's modality (originality.ai, pangram.com, read 8 November 2026).
Do either train on the text you submit?
Pangram says no: its detection is trained on a proprietary dataset that excludes user submissions. Originality.ai says scan history may train its models unless the user opts out (pangram.com/data-privacy, originality.ai/privacy-policy, read 8 November 2026).
Sources and revisions
Sources
- Pangrampangram.com
- Pangram pricingpangram.com/pricing
- Pangram data privacypangram.com/data-privacy
- Pangram aboutpangram.com/about-us
- Originality.aioriginality.ai
- Originality.ai pricingoriginality.ai/pricing
- Originality.ai privacy policyoriginality.ai/privacy-policy
- Originality.ai extensionsoriginality.ai/extensions
- Originality.ai 5.0 postoriginality.ai/blog/introducing-originali...
- Jabarian and Imas 2025 (read 9 November 2026)nber.org/papers/w34223
- Van Vlasselaer 2026 (read 9 November 2026)doi.org/10.1007/s40979-026-00226-w
- Liang et al. 2023arxiv.org/pdf/2304.02819v3
- Walters 2023degruyterbrill.com/document/doi/10.1515/o...
- Dugan et al. 2024 (RAID)arxiv.org/pdf/2405.07940
- Orenstrakh et al. 2023 (read 9 November 2026)arxiv.org/abs/2307.07411
- Pegoraro et al. 2023 (read 9 November 2026)arxiv.org/abs/2304.01487
