Pangram vs Copyleaks AI detector: two text detectors with one shared study, on the same 14 fields

AIDI-C-13

Pangram and Copyleaks are the two cards in the index that both publish an API price per word and both carry a study-measured false positive rate of zero, and they share one study record: Van Vlasselaer 2026 ran both on the same 40 fully AI-generated papers, 40 hybrid papers, 40 humanised papers and 40 human papers, and classified all 40 human papers correctly with each, while Pangram reached 97.5% inclusive accuracy on the fully AI set and Copyleaks scored 75% of the same 40 papers under 20% (DOI 10.1007/s40979-026-00226-w, Tables 4 and 5, read 9 November 2026). Copyleaks carries 9 study records to Pangram's 4, the two false positive zeros rest on different bases (0 of 42 papers in Walters 2023 for Copyleaks, 0.0000 of 1,992 passages in Jabarian and Imas 2025 for Pangram), and the vendors' own pages state 1 in 10,000 and 0.30% (pangram.com, copyleaks.com/ai-detector/testing-methodology, read 8 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.

Updated Fields checked 8 November 2026

Short answer

Pangram states 24 languages, 20 free checks a day after sign-up, plans from $20 a month and a text API at $0.05 per 100 words, with text stored for as long as the account is active and a zero data retention option for the enterprise API; Copyleaks states 30 languages (11 named), a free scan of up to 25,000 characters, plans from $16.99 a month with a free account unlocking more credits, an API sold through the same credit plans, $16.99 a month for 25,000 words, and a privacy policy that allows model training unless a direct customer opts out (pangram.com/pricing, pangram.com/data-privacy, copyleaks.com/pricing, copyleaks.com/privacy-policy, read 8 November 2026). Both carry a measured zero on human text, Pangram on 1,992 passages in Jabarian and Imas 2025 and Copyleaks on 42 papers in Walters 2023, and the one study that ran both, Van Vlasselaer 2026, found Pangram at 97.5% inclusive accuracy on 40 fully AI-generated papers where Copyleaks scored 75% of them under 20% (Jabarian and Imas 2025; Walters 2023; Van Vlasselaer 2026). Copyleaks also carries the index's highest false accusation count in one study, 5 of 10 human samples in Perkins 2024 (Perkins 2024). No detector score should be the only evidence in a misconduct, hiring or publishing decision.

How this comparison is built

Pangram vs Copyleaks AI detector: two text detectors with one shared study, on the same 14 fields: register plate with the page title and its register ID
The comparison plate: register ID and title, drawn from the index data.

How do Pangram and Copyleaks compare on the 14 fields?

Pangram vs Copyleaks: the 14 fields side by side
FieldPangramCopyleaks
1. Name, vendor, official domainPangram, Pangram Labs, pangram.comCopyleaks AI Detector, Copyleaks, copyleaks.com
2. Modalities detectedtext, image, videotext, code, image, video
3. Generators coveredOpenAI, Anthropic, Google, Meta, DeepSeek, xAI and 22 named models from Claude Opus 5 to GLM 5.2; "26 models benchmarked"GPT-5, GPT-4, ChatGPT, Claude, Gemini, DeepSeek, Llama, Bloom, Rytr, Jasper; image: OpenAI, Google, Midjourney, Stability AI, Black Forest Labs
4. Languages supported24, all named, from English and Arabic to Vietnamese30 stated, 11 named: English, French, German, Italian, Portuguese, Spanish, Japanese, Chinese (Simplified and Traditional), Russian, Dutch
5. Input limitsWords per text check not published; PDF, DOCX, RTF, up to 100 files at a time; API 5 queries a second, bulk 1,000 billable unitsMinimum 255 characters and a maximum of 2,000 pages on the web platform; 350 to 25,000 characters in the browser extension; 1 credit covers 250 words or 1 image
6. Free tier20 free credits daily after signing up for the free planFree scan of up to 25,000 characters; a limited number of free credits for new users, count not published; a free account unlocks more monthly credits
7. Paid plansIndividual $20 a month (300,000 words, 100 image scans, $10 of API credits); Professional $65 a month; Team $40 a month for 2 seatsPersonal $16.99 a month or $13.99 a month billed annually (100 credits, 25,000 words); Pro $99.99 a month or $74.99 billed annually (25 seats, 1,000 credits); Education and Enterprise custom
8. APIdocs.pangram.com: text $0.05 per 100 words, images from $0.05, video $0.05 a second; bulk text 20% offdocs.copyleaks.com: priced through the credit plans, 1 credit per 250 words
9. IntegrationsChrome and Firefox extensions, Google Docs, Gmail, Google Classroom, Canvas, Moodle, Brightspace (D2L), MCP serverCanvas, D2L, Moodle, Blackboard, Schoology, Edsby, Sakai, Google Classroom, browser extension (Chrome, Edge, Firefox), Google Docs add-on, WordPress add-on, MCP access
10. Vendor accuracy and false positive claims"accuracy of over 99%"; "accurate on 99.6% of these AI-generated samples"; "1 in 10,000 False positive rate" on "tens of millions" of public documentsMethodology page (model V11, test date 10 September 2026): human-only total accuracy 0.9970, 923 of 326,120 human texts flagged, 0.30% false positives at the default sensitivity; product page: over 99% (accuracy) and ".03% false positive rate"
11. Independent studies4 records: Jabarian and Imas 2025, Van Vlasselaer 2026, DAMAGE 2025 (research models), Epoch AI 20269 records: Walters 2023, Elkhatat 2023, Perkins 2024, Chaka 2023, Orenstrakh 2023, Van Vlasselaer 2026, Pegoraro 2023, Kar 2024, Alshammari and Rao 2025
12. Data retention and privacy statementSubmitted text stored "for as long as your account is active", deletable from the history tab; not used to train models; "Zero data retention options for enterprise API"Privacy Policy (24 August 2026): data may be used for "training Copyleaks' models", opt-out by contacting support; API FAQ: API content "is not stored, shared, or used to train public AI models"; a Shared Data Hub of user-submitted documents
13. Launch year and ownership2023; Pangram Labs, founded in Brooklyn by Max Spero and Bradley Emi; no parent statedNot stated; Copyleaks "together with its affiliated companies"; no parent stated
14. Rankings the card appears inIndependently tested, false positive rate, languages, studies by content type, adversarial robustness, false negative rate, claim vs evidence gap, platform integrations, file types, API pricing, LMS coverage, free checks per day, generators covered, non-English evidenceIndependently tested, false positive rate, languages, studies by content type, adversarial robustness, false negative rate, claim vs evidence gap, platform integrations, file types, minimum input, API pricing, LMS coverage, generators covered, non-English evidence

Which studies tested Pangram or Copyleaks?

One study tested both. Van Vlasselaer 2026 ran four detectors on 40 papers in each of four classes, fully AI-generated by ChatGPT Deep Research, hybrid, humanised and human-written, and printed each tool's score bands and accuracy per class (DOI 10.1007/s40979-026-00226-w, Tables 4 and 5, read 9 November 2026); the authors then ran Pangram alone on 1,163 theses. Pangram's other three records are Jabarian and Imas 2025, 1,992 human and 1,992 AI passages scored through the API at a 0.5 threshold (DOI 10.3386/w34223, Tables 3, 4 and 6, read 9 November 2026), Epoch AI 2026, 495 pre-2022 human passages and 297 AI passages per prompt on Pangram 3.3.2 (epoch.ai, read 9 November 2026), and DAMAGE 2025, where the figures belong to Pangram Labs' research models and the paper does not state whether DAMAGE is the shipping product (arXiv 2501.03437, Tables 3 and 7, read 9 November 2026). Copyleaks' other eight records run from Walters 2023, 126 documents (De Gruyter 10.1515/opis-2022-0158, Table 4, read 8 November 2026), to Alshammari and Rao 2025, 49 DeepSeek-V3 answers against 49 human answers (arXiv 2507.17944, Tables III and V, read 9 November 2026). Neither tool was in Weber-Wulff 2023 (Weber-Wulff 2023).

SourcePangramCopyleaks
Van Vlasselaer 2026, 40 fully AI-generated papers (Table 4)65% strict accuracy, 97.5% inclusive accuracy75% of the 40 scored under 20%, 25% between 20% and 40%
Van Vlasselaer 2026, 40 hybrid papers (Table 5)37 of 40 within their known AI range30.0% accuracy
Van Vlasselaer 2026, 40 humanised papers (Table 5)37 of 40 within their known AI range22.5% accuracy
Van Vlasselaer 2026, 40 human-written papersAll 40 classified correctlyAll 40 classified correctly
Van Vlasselaer 2026, 1,163 theses run on Pangram alone (Table 6)45.5% flaggedNot run
Jabarian and Imas 2025, 1,992 human passages at threshold 0.5 (Table 3)False positive rate 0.0000; at most 0.0075 per genre at Youden-optimised thresholds (Table 6)Not tested
Jabarian and Imas 2025, 1,992 AI passages per model (Table 4)False negative rate at most 0.0382; 0.0131 on GPT-4.1 at threshold 0.5; 0.0000 to 0.0500 after StealthGPT rewrites (Table B.4)Not tested
Epoch AI 2026, Pangram 3.3.2 at its default verdict0 of 495 pre-2022 human passages flagged; 0 of 297 plain-prompt AI passages missed; 30 of 297 (10.10%) missed when written in a named author's styleNot tested
DAMAGE 2025, Pangram Labs' research models (Tables 3 and 7)DAMAGE model: 100% true positive rate on AI essays and 98.26% after humanizing at 5% FPR; 3.40% false positive rate at the default threshold; not stated to be the shipping productNot tested
Walters 2023, 126 documents (Table 4)Not tested100% correct: 0 of 42 human papers flagged, 100% on 42 GPT-3.5 and 42 GPT-4 papers
Elkhatat 2023, 15 paragraphs per class (Table 4)Not tested15 of 15 GPT-3.5 and 14 of 15 GPT-4 paragraphs flagged; 1 of 5 human controls flagged
Perkins 2024, 104 AI and 10 human samples (Tables 7, 8 and 10)Not tested64.8% baseline mean accuracy, the highest of seven; 73.9% on unmanipulated AI text falling to 58.7% after adversarial techniques; 5 of 10 human samples falsely accused, the highest; 41 of 104 AI samples undetected, the lowest
Chaka 2023, 21 English chatbot responses and 6 per translationNot tested15 of 21 English responses classified as AI, rank 1 of 5; 3 of 6 German, 5 of 6 French and 6 of 6 Spanish translations
Orenstrakh 2023, 114 human and 20 ChatGPT submissions (Tables 6, 8, 9 and 12)Not tested99.12% on human and 95.00% on ChatGPT English submissions, rank 1 of 8, 1 false positive in 114; 100% to 50% on 10 QuillBot-paraphrased texts; all 10 Spanish human papers flagged as AI
Pegoraro 2023, ChatGPT and human Q&A benchmark (Table I)Not testedTrue positive rate 22.9%, true negative rate 92.1%
Kar 2024, one ChatGPT 3.5 article and three paraphrases (Table 1)Not tested100% AI on the original; 0 after a QuillBot paraphrase; 100 after Grammarly and ChatGPT paraphrases
Alshammari and Rao 2025, 49 answers per set (Tables III and V)Not testedMean AI score 98.8% on DeepSeek-V3 answers; mean human score 100% on 49 human answers; 99.7% on DeepThink text, 71% after a humanized paraphrase
Vendor statement, pangram.com (read 8 November 2026)"1 in 10,000 False positive rate", "calculated on an aggregate of public datasets containing tens of millions of written documents"
Vendor statement, copyleaks.com (read 8 November 2026)0.30% false positives and 0.64% false negatives at the Balanced default (methodology page, model V11, 10 September 2026); ".03% false positive rate" on the product page

Results as each study reports them: DOI 10.1007/s40979-026-00226-w, DOI 10.3386/w34223, epoch.ai, arXiv 2501.03437, De Gruyter 10.1515/opis-2022-0158, DOI 10.1007/s40979-023-00140-5, DOI 10.1186/s41239-024-00487-w, DOI 10.37074/jalt.2023.6.2.12, arXiv 2307.07411, arXiv 2304.01487, DOI 10.1177/02537176241247934, arXiv 2507.17944, read 8 and 9 November 2026. "Not tested" means the paper did not run that tool. The false positive rate ranking sorts Pangram on the Jabarian and Imas 2025 figure and Copyleaks on the Walters 2023 figure; the adversarial robustness ranking carries Copyleaks' 58.7% after adversarial techniques from Perkins 2024 and Pangram's DAMAGE figures stay in prose as research models.

What do the two cost over 12 months?

PlanMonthly as statedAnnual as statedInclusions as statedChecked on
Pangram free plan$020 free credits daily after sign-up, including AI assistance detection8 November 2026
Pangram Individual$20 per month; 7-day free trial"Save $60, annually": $180 at the stated saving300,000 words a month, 100 image scans, similarity detection, MCP, Google Docs and Gmail integration, $10 of monthly API credits8 November 2026
Pangram Professional$65 per month"Save $240, annually": $540 at the stated saving5x the words and image scans of Individual, $200 of API credits a month8 November 2026
Pangram Team$40 per month for 2 seats ($20 a seat)Not shownTeam seats at $20 each8 November 2026
Pangram API$0.05 per 100 words of text (Pangram 4.1); Image 1.0 $0.25 and Image Flash $0.05 per image; video $0.05 a secondMeteredBulk text API 20% off8 November 2026
Copyleaks free scan$0Up to 25,000 characters per scan; a limited number of free credits for new users, count not published8 November 2026
Copyleaks Personal$16.99 per month billed monthly; $13.99 per month billed annually$167.88 at the stated annual monthly rate times twelve100 unified credits a month: up to 25,000 words or 100 images; AI and plagiarism detection for text, AI detection for images8 November 2026
Copyleaks Pro$99.99 per month billed monthly; $74.99 per month billed annually$899.88 at the stated annual monthly rate times twelve25 user seats; 1,000 unified credits a month: up to 250,000 words or 1,000 images8 November 2026
Copyleaks Education and EnterpriseCustomCustom"Custom" on the pricing page8 November 2026

Prices in USD as shown on pangram.com/pricing and copyleaks.com/pricing, checked 8 November 2026. Pangram's annual figures are the index's subtraction of the stated saving from twelve monthly payments; Copyleaks' are the index's multiplication of the stated annual monthly rate by twelve, because neither page prints a yearly total. On words alone, Pangram Individual's 300,000 words a month for $20 and Copyleaks Personal's 25,000 words a month for $16.99 are the two entry allowances; through the APIs, Pangram's $0.05 per 100 words is $0.50 per 1,000 words and Copyleaks' Personal monthly plan is $0.68 per 1,000 words at the full allowance, the two figures the API pricing ranking carries.

One of the comparisons in the compare hub. The cards are Pangram and Copyleaks, and the home board answers which AI detectors exist and who has tested them? An editor or integrity officer choosing between the two API-first text detectors, each with a zero on human text in at least one study, is the reader this pair is written for; the false positive rate ranking sorts both on the same field and the API pricing ranking converts both prices to the same unit.

Which card fits which reader?

  • Choose Pangram ifyou want the larger measured human base behind the zero: 0.0000 of 1,992 passages in Jabarian and Imas 2025 and 0 of 495 in Epoch AI 2026 against 0 of 42 in Walters 2023 for Copyleaks (DOI 10.3386/w34223, Table 3; epoch.ai; De Gruyter, Table 4).
  • Choose Pangram ifyour text comes from ChatGPT Deep Research or has been humanised: 97.5% inclusive accuracy on the fully AI set and 37 of 40 humanised papers in range, against 75% of the fully AI set under 20% and 22.5% on humanised papers for Copyleaks in the one study that ran both (DOI 10.1007/s40979-026-00226-w, Tables 4 and 5, read 9 November 2026).
  • Choose Copyleaks ifyou want the longer outside record: 9 study records from 2023 to 2025 against Pangram's 4, including 100% of 126 documents in Walters 2023 and rank 1 of 8 in Orenstrakh 2023 (Walters 2023; Orenstrakh 2023).
  • Choose Copyleaks ifthe check must run inside Blackboard, Schoology, Edsby or Sakai: it names eight LMS platforms where Pangram names Google Classroom, Canvas, Moodle and Brightspace (copyleaks.com/ai-content-detector, pangram.com, read 8 November 2026).
  • Choose Pangram ifyou want a per-word API price without a plan: $0.05 per 100 words, metered, against Copyleaks' credit plans (pangram.com/pricing, copyleaks.com/pricing, read 8 November 2026).
  • Choose Pangram ifyou need the text kept out of model training by default: the data privacy page states that detection is built on a proprietary dataset that does not include user submissions, where Copyleaks' policy allows training unless a direct customer opts out (pangram.com/data-privacy, copyleaks.com/privacy-policy, read 8 November 2026).
  • Choose Copyleaks ifyou also need source code checked: its modalities include code, through the Codeleaks product with its own card (Codeleaks card); Pangram states text, image and video.
  • Choose Pangram ifyou check text in Czech, Greek, Hindi, Hungarian, Persian, Polish, Romanian, Swedish, Turkish, Ukrainian, Urdu or Vietnamese: Pangram names all 24 of its languages and Copyleaks names 11 of its 30 (pangram.com, copyleaks.com/ai-content-detector, 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 Copyleaks
1: Van Vlasselaer 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
Read the Pangram card

Copyleaks

Register ID
AIDI-D-0003
Study records
9 in the index: Walters 2023, Elkhatat 2023, Perkins 2024, Chaka 2023, Orenstrakh 2023, Van Vlasselaer 2026, Pegoraro 2023, Kar 2024, Alshammari and Rao 2025
Shared with Pangram
1: Van Vlasselaer 2026
Vendor accuracy claim
Methodology page (test date September 10, 2026; publish date September 28, 2026; model V11): QA human-only total accuracy 0.9970 (923 of 326,120 human texts flagged as AI). Product page: accuracy stated as over 99%, with an industry-low .03% false positive rate; Copyleaks leads the field in trusted detection.
Free tier
Free for up to 25,000 characters per scan; every new user receives a limited number of free credits (count not published); a free account unlocks more monthly credits
Languages
30 stated
Read the Copyleaks card

How this comparison is built

Cards compared
Pangram and Copyleaks
Fields
The 14 fields of every card, in card order
Shared study records
1: Van Vlasselaer 2026
Fields checked
8 November 2026
Records in the studies index
35 records

Read the full methodology

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. One record ran both tools, Van Vlasselaer 2026, so the evidence table opens with its five rows and lists the other twelve records with "Not tested" in the column of the tool the paper did not run; the DAMAGE 2025 row is labelled as Pangram Labs' research models because the paper does not say whether DAMAGE is the shipping product, and each vendor's own statement sits in a labelled row. Copyleaks' two vendor false positive figures, .03% on the product page and 0.30% on the methodology page, are both printed because the index records what the pages say. The rules, the recheck schedule (next pass 8 December 2026) and the correction route are on the methodology page; the claim vs evidence gap ranking pairs Pangram's over 99% with 97.5% and Copyleaks' over 99% with 100%, the only claim in the index that undershoots its evidence by a point.

Where each figure comes from

Values as each vendor states them on pangram.com, pangram.com/pricing, pangram.com/data-privacy, copyleaks.com/ai-content-detector, copyleaks.com/pricing, copyleaks.com/ai-detector/testing-methodology and copyleaks.com/privacy-policy, all read 8 November 2026. Copyleaks' product page and its methodology page print two different false positive figures, .03% and 0.30%, and the card records both; Pangram's 1 in 10,000 is stated on an aggregate of public datasets with no dataset named.

Questions about Pangram and Copyleaks

Did any study test Pangram and Copyleaks on the same text?

Yes, one: Van Vlasselaer 2026 ran both on the same 160 papers and classified all 40 human papers correctly with each; Pangram reached 97.5% inclusive accuracy on the 40 fully AI-generated papers and Copyleaks scored 75% of them under 20% (DOI 10.1007/s40979-026-00226-w, Tables 4 and 5, read 9 November 2026).

Which has the lower measured false positive rate?

Both carry a zero: Pangram's 0.0000 of 1,992 human passages in Jabarian and Imas 2025 (DOI 10.3386/w34223, Table 3) and Copyleaks' 0 of 42 human papers in Walters 2023 (De Gruyter 10.1515/opis-2022-0158, Table 4), read 8 and 9 November 2026. Copyleaks also recorded 5 false accusations in 10 human samples in Perkins 2024 and 1 in 114 in Orenstrakh 2023; the ranking takes the lowest printed rate per card.

Which has more study records?

Copyleaks, with 9 to Pangram's 4, by the study lists on the two cards read 10 November 2026 (Copyleaks card, Pangram card); Copyleaks' records start in March 2023 and Pangram's in 2025.

Which is cheaper per word through the API?

Pangram states $0.05 per 100 words, $0.50 per 1,000 words, metered with no plan; Copyleaks sells credits, 1 per 250 words, at $16.99 a month for 100 credits, $0.68 per 1,000 words at the full allowance, the figure the API pricing ranking converts (pangram.com/pricing, copyleaks.com/pricing, read 8 November 2026). A reader who uses fewer than 25,000 words a month pays more per word on the plan, and the annual rate of $13.99 lowers it.

Does either train on the text you submit?

Pangram's data privacy page says its detection is "based off of a large, proprietary dataset that doesn't include user submitted content"; Copyleaks' Privacy Policy of 24 August 2026 allows use "including by training Copyleaks' models" with an opt-out for direct customers through support, while its API FAQ says API content is not used to train public AI models (pangram.com/data-privacy, copyleaks.com/privacy-policy, read 8 November 2026).

Which supports more languages?

Copyleaks states 30 and names 11; Pangram states and names 24 (copyleaks.com/ai-content-detector, pangram.com, read 8 November 2026). The one non-English result for either is Copyleaks': 3 of 6, 5 of 6 and 6 of 6 translated ChatGPT responses flagged in Chaka 2023, and all 10 Spanish human papers flagged in Orenstrakh 2023 (Chaka 2023; Orenstrakh 2023).

What happened to each after paraphrasing?

Copyleaks fell from 73.9% to 58.7% on AI text after adversarial techniques in Perkins 2024 and from 100% to 50% on 10 QuillBot-paraphrased texts in Orenstrakh 2023; Pangram scored 37 of 40 humanised papers within their known AI range in Van Vlasselaer 2026 (Perkins 2024; Orenstrakh 2023; Van Vlasselaer 2026).

Sources and revisions

Sources

  1. Pangram home page (languages, generators, claims)pangram.com
  2. Pangram pricingpangram.com/pricing
  3. Pangram data privacypangram.com/data-privacy
  4. Pangram API docsdocs.pangram.com/quickstart
  5. Copyleaks AI Detectorcopyleaks.com/ai-content-detector
  6. Copyleaks pricingcopyleaks.com/pricing
  7. Copyleaks testing methodologycopyleaks.com/ai-detector/testing-methodo...
  8. Copyleaks privacy policycopyleaks.com/privacy-policy
  9. Van Vlasselaer 2026 (read 9 November 2026)doi.org/10.1007/s40979-026-00226-w
  10. Jabarian and Imas 2025 (read 9 November 2026)doi.org/10.3386/w34223
  11. Epoch AI 2026 (read 9 November 2026)epoch.ai/data-insights/ai-detectors-false...
  12. DAMAGE 2025 (read 9 November 2026)arxiv.org/pdf/2501.03437
  13. Walters 2023degruyterbrill.com/document/doi/10.1515/o...
  14. Elkhatat 2023doi.org/10.1007/s40979-023-00140-5
  15. Perkins 2024 (read 9 November 2026)doi.org/10.1186/s41239-024-00487-w
  16. Chaka 2023 (read 9 November 2026)doi.org/10.37074/jalt.2023.6.2.12
  17. Orenstrakh 2023 (read 9 November 2026)arxiv.org/pdf/2307.07411
  18. Pegoraro 2023 (read 9 November 2026)arxiv.org/abs/2304.01487
  19. Kar 2024 (read 9 November 2026)doi.org/10.1177/02537176241247934
  20. Alshammari and Rao 2025 (read 9 November 2026)arxiv.org/abs/2507.17944

Revision log

  1. Comparison created at index version 2026

Corrections log

Related reading