Converged.AI × BrandInsightAI
BrandInsightAI Insights
Insights & Analysis

AI search, measured.

Which websites do AI models actually cite, and how is that changing? Living leaderboards and regular analysis from the BrandInsightAI dataset.

Behind every figure: 264,296 citations · 55,285 model responses in the rolling window · measured, never estimated

The atlas · 8 weeks to 2026-09-14

ChatGPT is the most authority-seeking of the 7 models citing this week — 17% of what it cites is government or educational.

One ring per model, same slice order and same ink scale on every ring, from every citation every project produced in the window — not a survey. Hover a slice for the share.

ChatGPT is the most authority-seeking of the 7 models citing this week — 17% of what it cites is government or educational. Claude · Personal 2%Claude · Brand 33%brandClaude · Government/Civic 0%Claude · E-commerce 7%Claude · News Publisher 14%newsClaude · Educational/Non-Profit 6%Claude · Platform 12%Claude · Utility/Tool 6%Claude · Other/Unclassified 20%otherClaude Perplexity · Personal 2%Perplexity · Brand 29%brandPerplexity · Government/Civic 0%Perplexity · E-commerce 6%Perplexity · News Publisher 15%newsPerplexity · Educational/Non-Profit 5%Perplexity · Platform 13%platformPerplexity · Utility/Tool 7%Perplexity · Other/Unclassified 22%otherPerplexity Gemini · Personal 2%Gemini · Brand 33%brandGemini · Government/Civic 0%Gemini · E-commerce 11%Gemini · News Publisher 10%Gemini · Educational/Non-Profit 8%Gemini · Platform 12%Gemini · Utility/Tool 5%Gemini · Other/Unclassified 18%otherGemini ChatGPT · Personal 1%ChatGPT · Brand 40%brandChatGPT · Government/Civic 1%ChatGPT · E-commerce 11%ChatGPT · News Publisher 9%ChatGPT · Educational/Non-Profit 16%eduChatGPT · Platform 7%ChatGPT · Utility/Tool 4%ChatGPT · Other/Unclassified 12%ChatGPT Grok · Personal 1%Grok · Brand 36%brandGrok · Government/Civic 1%Grok · E-commerce 7%Grok · News Publisher 27%newsGrok · Educational/Non-Profit 5%Grok · Platform 8%Grok · Utility/Tool 5%Grok · Other/Unclassified 9%Grok no citationsyetDeepSeek Google AIO · Personal 1%Google AIO · Brand 32%brandGoogle AIO · Government/Civic 0%Google AIO · E-commerce 9%Google AIO · News Publisher 9%Google AIO · Educational/Non-Profit 11%Google AIO · Platform 13%platformGoogle AIO · Utility/Tool 9%Google AIO · Other/Unclassified 14%otherGoogle AIO Google AI Mode · Personal 1%Google AI Mode · Brand 33%brandGoogle AI Mode · Government/Civic 0%Google AI Mode · E-commerce 9%Google AI Mode · News Publisher 10%Google AI Mode · Educational/Non-Profit 11%Google AI Mode · Platform 11%Google AI Mode · Utility/Tool 10%Google AI Mode · Other/Unclassified 14%otherGoogle AI Mode no citationsyetMeta AI
  1. ClaudeBrand first — 33% of its citations
  2. PerplexityBrand first — 29% of its citations
  3. GeminiBrand first — 33% of its citations
  4. ChatGPTBrand first — 40% of its citations
  5. GrokBrand first — 36% of its citations
  6. DeepSeekno citations yet
  7. Google AIOBrand first — 32% of its citations
  8. Google AI ModeBrand first — 33% of its citations
  9. Meta AIno citations yet
The shares behind the rings
Source typeClaudePerplexityGeminiChatGPTGrokDeepSeekGoogle AIOGoogle AI ModeMeta AI
Personal2%2%2%1%1%1%1%
Brand33%29%33%40%36%32%33%
Government/Civic0%0%0%1%1%0%0%
E-commerce7%6%11%11%7%9%9%
News Publisher14%15%10%9%27%9%10%
Educational/Non-Profit6%5%8%16%5%11%11%
Platform12%13%12%7%8%13%11%
Utility/Tool6%7%5%4%5%9%10%
Other/Unclassified20%22%18%12%9%14%14%

Pick your view

Everything is measured from real AI answers — six ways in, depending on the question you're asking. They interlock: the Category Report tells the story the Category Index ranks, the GEO Index and Source Index show which websites feed those answers, and the Methodology governs every figure on all of them. The bar at the top of every page carries you between them.

Live leaderboards

Auto-updated reference pages: the current state of AI citation behaviour, refreshed hourly.

Analysis

Weekly movers, model deep-dives, and the evidence on what makes content citable. Every edition is generated from the same warehouse the boards above render, and reviewed before it publishes.

Citation Movers · 8 Sept 2026
AI Citation Movers · week commencing 31 Aug 2026
gov.uk gained 0.5pp AI citation share to 0.7%. Measured across 127,742 citations and 26,758 model responses for the week commencing 31 Aug.
Model vs Model · 3 Sept 2026
Perplexity vs Google AIO: Where They Get Their Answers · August 2026
Two AI models, 114,314 citations in August 2026 — and clearly different reading habits.
Citation Movers · 3 Sept 2026
AI Citation Movers · week commencing 24 Aug 2026
Measured across 80,661 citations and 16,083 model responses in the week commencing 24 Aug, google.com led weekly AI citation gainers.
Model vs Model · 29 Aug 2026
Gemini vs ChatGPT: Where They Get Their Answers · July 2026
Two AI models, 63,430 citations in July 2026, and clearly different reading habits.
Model vs Model · 27 Aug 2026
Claude vs ChatGPT: Where They Get Their Answers · July 2026
An analysis of 34,876 Claude citations and 12,478 ChatGPT citations reveals stark differences in source selection alongside moderate rank agreement.
Model vs Model · 26 Aug 2026
Grok vs Google AI Mode: Where They Get Their Answers · May 2026
Grok and Google AI Mode reveal distinct citation habits across thousands of analysed responses, showing clear divergence in their source mixes.
Model vs Model · 26 Aug 2026
Google AIO vs Google AI Mode: Where They Get Their Answers · June 2026
An analysis of citation patterns reveals how Google AIO and Google AI Mode diverge in their reliance on specific domains and platforms.
AI Citations Report · 26 Aug 2026
The AI Citations Report · Q2 2026
Across 623,900 citations over 13 weeks, BrandInsightAI's AI Source Index tracks the exact visibility patterns of top domains across 7 AI models.
What Gets Cited · 26 Aug 2026
What Gets Cited by AI · July 2026 Benchmarks
Feature lifts, winning formats, title language and the anatomy of quoted passages, benchmarks across 16 content audits spanning 6 projects.
Model vs Model · 26 Aug 2026
Claude vs Perplexity: Where They Get Their Answers · July 2026
An analysis of July 2026 data reveals starkly different reading habits between Claude and Perplexity, from video platforms to encyclopaedias.