Not sure where to start? Take a guided tour.
AI Site Grader
AI Query Scanner
llms.txt GeneratorWhy do models disagree so much even on common categories?
Because they prioritize different evidence sets, training priors, and retrieval habits. The output looks like one market, but the study shows 8 distinct recommendation systems with only partial overlap.
The key numbers.
- Average agreement
- Mean cross-model agreement rate.
- 43.3%
- High divergence rate
- Prompts in the 0-25% agreement bucket.
- 14.6%
- Models analyzed
- OpenAI, Anthropic, Gemini, Grok, Deepseek, Meta, Perplexity, and Google AI Overviews.
- 8
The disagreement is not random error. It is a structural feature of cross-model visibility.
The evidence.
| Measure | Result |
|---|---|
| Average agreement Mean cross-model agreement rate. | 43.3% |
| High divergence rate Prompts in the 0-25% agreement bucket. | 14.6% |
| Models analyzed OpenAI, Anthropic, Gemini, Grok, Deepseek, Meta, Perplexity, and Google AI Overviews. | 8 |
Population and method
Built from 797,644 valid comparisons across 44,088 reports and 8 models, covering 6,439,133 model responses in the observed window.
Where the evidence stops
Agreement is measured across recommendation outputs, not across hidden reasoning or retrieval context.
Average agreement can hide large differences between query classes and model pairs.
The study measures overlap, not which answer was objectively “right”.
What to do with this
- Track visibility across multiple models instead of using one platform as a proxy for the whole market.
- Prioritize query classes where disagreement is highest because that is where share can move fastest.
- Treat consensus as a benchmark, but treat divergence as the operating reality.
This answer matters because it turns a study finding into an operating rule teams can use when they decide what to publish, refresh, or measure next.
Questions about this answer.
Cite this answer.
Source study: Model Divergence“Because they prioritize different evidence sets, training priors, and retrieval habits. The output looks like one market, but the study shows 8 distinct recommendation systems with only partial overlap.”
Mack Grenfell · Why do models disagree so much even on common categories?
Same Question, Different AI, Different Answers · Trakkr Research · Updated 11 March 2026