What was counted?

Shares of each engine's classified citations (citations whose source role the Index classifies), across the Index's 1,255 buyer questions, cumulative from May 10.

What are the figures?

ChatGPT (gpt-4.1-mini, API): 39.9% (4,361 of 10,938 classified citations).

ChatGPT (gpt-4.1-mini, API)39.9% (4,361 of 10,938 classified citations)
Google AI Overviews16.0% (509 of 3,174 classified citations)
Editorial rank among ChatGPT source roles1
Editorial rank among Google AI Overviews source roles3
Ratio2.49 to 1
Gap23.8 percentage points
Index answer runs observed40,964

How has this changed across releases?

The figure was 40.0% vs 16.0% on the November 17, 2026 Index release and is 39.9% vs 16.0% on the November 18, 2026 Index release.

Index releaseFigureWhat changed
November 18, 2026 Index release39.9% vs 16.0%ChatGPT (gpt-4.1-mini, API): 40.0% (4,328 of 10,825 classified citations) to 39.9% (4,361 of 10,938 classified citations). Google AI Overviews: 16.0% (506 of 3,158 classified citations) to 16.0% (509 of 3,174 classified citations). Ratio: 2.50 to 1 to 2.49 to 1. 17 more figures moved.
November 17, 2026 Index release40.0% vs 16.0%ChatGPT (gpt-4.1-mini, API): 39.9% (4,276 of 10,723 classified citations) to 40.0% (4,328 of 10,825 classified citations). Google AI Overviews: 16.1% (505 of 3,139 classified citations) to 16.0% (506 of 3,158 classified citations). Ratio: 2.48 to 1 to 2.50 to 1. 19 more figures moved.
November 16, 2026 Index release39.9% vs 16.1%ChatGPT (gpt-4.1-mini, API): 40.2% (4,039 of 10,039 classified citations) to 39.9% (4,276 of 10,723 classified citations). Google AI Overviews: 16.3% (488 of 2,995 classified citations) to 16.1% (505 of 3,139 classified citations). Ratio: 2.47 to 1 to 2.48 to 1. 20 more figures moved.
November 9, 2026 Index release40.2% vs 16.3%ChatGPT (gpt-4.1-mini, API): 40.4% (3,672 of 9,098 classified citations) to 40.2% (4,039 of 10,039 classified citations). Google AI Overviews: 16.3% (469 of 2,877 classified citations) to 16.3% (488 of 2,995 classified citations). Ratio: 2.48 to 1 to 2.47 to 1. 18 more figures moved.
November 2, 2026 Index release40.4% vs 16.3%ChatGPT (gpt-4.1-mini, API): 40.9% (3,267 of 7,990 classified citations) to 40.4% (3,672 of 9,098 classified citations). Google AI Overviews: 16.6% (457 of 2,756 classified citations) to 16.3% (469 of 2,877 classified citations). Ratio: 2.47 to 1 to 2.48 to 1. 20 more figures moved.
October 26, 2026 Index release40.9% vs 16.6%First release in this series.

What is each engine's editorial share?

ChatGPT (gpt-4.1-mini, API) sends 39.9% of its classified citations to editorial publications. Google AI Overviews sends 16.0%. The ChatGPT share pools every model era in the window.

EngineEditorial citationsClassified citationsEditorial share
ChatGPT (gpt-4.1-mini, API)4,36110,93839.9%
Perplexity11,16134,65032.2%
Claude3,40412,12928.1%
Google AI Mode1,9589,04721.6%
Gemini1,6808,22720.4%
Google AI Overviews5093,17416.0%

How does each engine divide its classified citations?

Editorial publications rank 1 among ChatGPT source roles and 3 among Google AI Overviews source roles.

All 48 rows
EngineSource roleCitationsClassified citationsShareRank in engine
ChatGPTeditorial_media4,36110,93839.9%1
ChatGPTvendor_owned2,54810,93823.3%2
ChatGPTacademic_government1,17910,93810.8%3
ChatGPTplatform_search71310,9386.5%4
ChatGPTmarket_database70810,9386.5%5
ChatGPTanalyst_research59510,9385.4%6
ChatGPTcommunity_social44910,9384.1%7
ChatGPTwire_distribution38510,9383.5%8
Claudevendor_owned4,36112,12936.0%1
Claudeeditorial_media3,40412,12928.1%2
Claudeacademic_government1,33312,12911.0%3
Claudemarket_database1,25512,12910.3%4
Claudeanalyst_research1,00412,1298.3%5
Claudecommunity_social61212,1295.0%6
Claudewire_distribution10712,1290.9%7
Claudeplatform_search5312,1290.4%8
Geminivendor_owned2,8738,22734.9%1
Geminieditorial_media1,6808,22720.4%2
Geminicommunity_social1,0888,22713.2%3
Geminiacademic_government8158,2279.9%4
Geminianalyst_research6658,2278.1%5
Geminimarket_database6388,2277.8%6
Geminiplatform_search4188,2275.1%7
Geminiwire_distribution508,2270.6%8
Google AI Modevendor_owned2,7809,04730.7%1
Google AI Modeeditorial_media1,9589,04721.6%2
Google AI Modecommunity_social1,2989,04714.3%3
Google AI Modeanalyst_research1,1349,04712.5%4
Google AI Modemarket_database8949,0479.9%5
Google AI Modeacademic_government6179,0476.8%6
Google AI Modeplatform_search2739,0473.0%7
Google AI Modewire_distribution939,0471.0%8
Google AI Overviewsvendor_owned8803,17427.7%1
Google AI Overviewscommunity_social7753,17424.4%2
Google AI Overviewseditorial_media5093,17416.0%3
Google AI Overviewsplatform_search3673,17411.6%4
Google AI Overviewsanalyst_research2443,1747.7%5
Google AI Overviewsacademic_government2113,1746.6%6
Google AI Overviewsmarket_database1723,1745.4%7
Google AI Overviewswire_distribution163,1740.5%8
Perplexityeditorial_media11,16134,65032.2%1
Perplexityvendor_owned9,97434,65028.8%2
Perplexityacademic_government3,85934,65011.1%3
Perplexitycommunity_social3,82834,65011.0%4
Perplexitymarket_database2,44234,6507.0%5
Perplexityanalyst_research2,42834,6507.0%6
Perplexityplatform_search81934,6502.4%7
Perplexitywire_distribution13934,6500.4%8

JSON holds every release of this table.

Which Briefs cite this finding?

These Briefs cite the Index release that was current when they were published. The history row for that release is the figure they used.

  • ChatGPT cites editorial 2.5 times as often, November 14, 2026. Cites the November 9, 2026 Index release (40.2% vs 16.3%). The next release moved: ChatGPT (gpt-4.1-mini, API): 40.2% (4,039 of 10,039 classified citations) to 39.9% (4,276 of 10,723 classified citations). Google AI Overviews: 16.3% (488 of 2,995 classified citations) to 16.1% (505 of 3,139 classified citations). Ratio: 2.47 to 1 to 2.48 to 1. 20 more figures moved.

Which records is this based on?

These are the records the script reads.

machinerelations.ai

How was this measured?

A script recomputes this page from the Index release whenever a new Index release publishes. Last changed November 18, 2026; last checked November 18, 2026.

Published by Machine Relations Research. Licensed CC BY 4.0.

Where is this measured?