The AI Disagreement Index
Where AI models
actually disagree.
When you ask GPT, Claude, Gemini and Grok the same question on Synero, the answers are independently graded for agreement: where the models agreed, where they split, and which one dissented. This AI model comparison index tracks the aggregates.
282
Answers graded
75/100
Average agreement
70%
Had a clear dissenter
Which AI disagrees most?
Dissent is counted when one model clearly departed from the rest of the group.
Agreement: Council vs Deep
| Models | Avg agreement | Chats |
|---|---|---|
| Council (balanced models) | 75/100 | 215 |
| Deep (flagship models) | 76/100 | 67 |
Methodology
After a multi-model answer completes (two or more models plus a combined answer), a grader model reads every response and scores substantive agreement from 0 to 100, lists the points all models share and the contested topics, and names a dissenting model when one clearly departed from the group. Individual questions and answers are never published here: this page shows aggregates only. Shared answers display their own per-run agreement report, published explicitly by the person who ran them.
See the disagreement on your own question
Ask your own question, see where the AIs agree, and share it as a link. Free to try, no card.
Learn how Synero works or see pricing.