Models
· how often eight AIs agree on the top brandof questions got the same top brand from every AI that answered. On the average question, 43% agree.
| Pair | Same top brand | Rank |
|---|---|---|
| Claude and DeepSeek | 35.0% | 1 of 28 |
| Claude and Grok | 34.6% | 2 of 28 |
| DeepSeek and Grok | 30.6% | 3 of 28 |
| Claude and ChatGPT | 27.1% | 4 of 28 |
| Claude and Gemini | 26.4% | 5 of 28 |
| DeepSeek and Gemini | 25.6% | 6 of 28 |
| DeepSeek and ChatGPT | 25.4% | 7 of 28 |
| Grok and Gemini | 23.7% | 8 of 28 |
| Grok and ChatGPT | 23.6% | 9 of 28 |
| Claude and Meta AI | 22.7% | 10 of 28 |
| DeepSeek and Meta AI | 21.5% | 11 of 28 |
| ChatGPT and Gemini | 20.7% | 12 of 28 |
| Grok and Meta AI | 20.6% | 13 of 28 |
| ChatGPT and AI Overviews | 20.1% | 14 of 28 |
| Claude and AI Overviews | 19.3% | 15 of 28 |
| DeepSeek and AI Overviews | 18.1% | 16 of 28 |
| AI Overviews and Perplexity | 17.3% | 17 of 28 |
| Grok and AI Overviews | 17.1% | 18 of 28 |
| Gemini and AI Overviews | 17.0% | 19 of 28 |
| ChatGPT and Perplexity | 16.6% | 20 of 28 |
| ChatGPT and Meta AI | 16.5% | 21 of 28 |
| Gemini and Meta AI | 15.9% | 22 of 28 |
| Claude and Perplexity | 15.1% | 23 of 28 |
| DeepSeek and Perplexity | 14.9% | 24 of 28 |
| Grok and Perplexity | 14.4% | 25 of 28 |
| Gemini and Perplexity | 12.2% | 26 of 28 |
| AI Overviews and Meta AI | 11.5% | 27 of 28 |
| Meta AI and Perplexity | 9.6% | 28 of 28 |
Claude and DeepSeek agree most. Even they pick the same top brand only 1 time in 3.
Each square is how often two AIs named the same top brand for the same question.
On most questions the AIs split, not agree
797,644 answered questions, by the share of AIs that picked the most common top brand.
| AIs that agreed | Answered questions | Share |
|---|---|---|
| Under 25% agree | 116,621 | 14.6% |
| 25 to 49% | 359,732 | 45.1% |
| 50 to 74% | 223,267 | 28.0% |
| 75 to 99% | 66,466 | 8.3% |
| All agree | 31,558 | 4.0% |
AI Overviews often names no top brand
Share of questions where each AI gave a ranked answer.
Comparison questions agree most
Average agreement on the top brand, by question type.
Line marks the 43% average. Best-of and general questions are 87% of the sample.
How we measure
We took 8,902 brand questions that Trakkr runs, asked 825,343 times between 12 Aug 2025 and 11 Mar 2026, and compared the top brand each AI named. Agreement on a question is the share of AIs that answered and picked the most common top brand. A pair agrees when both name the same top brand.
A question counts when at least two AIs gave a ranked answer; 797,644 did. The questions come from brands that use Trakkr, so they lean to the categories those brands track. This is a snapshot from 11 Mar 2026 and does not update. Example questions are not published.
Do AI models recommend the same brands?
Rarely all at once. Across 8,902 questions asked 825,343 times, only 4% of answered questions got the same top brand from every AI that answered; on the average question 43% of the AIs agreed. The closest pair, Claude and DeepSeek, named the same top brand 35% of the time.
Which AI models does Trakkr track?
This study compared eight: ChatGPT, Claude, Gemini, Google AI Overviews, Perplexity, Grok, DeepSeek and Meta AI. It is a snapshot from 11 Mar 2026 and does not update.
Why do AI models disagree on recommendations?
They are trained on different data, search different sources and weigh authority differently. This study measures how far apart they land, pair by pair and by question type; it does not test the reasons.