Independent model rankings
Everything,
ranked.
A public index of how leading language models judge the world. Same question, same field, repeated scores—and every disagreement left visible.
Model judgment
measured in public
measured in public
Published Sep 1, 2026Provider median = three samplesConsensus = median of provider mediansNo web search
Latest research
The Best Cities for Digital Nomads
The protocol
Repeatable by design.
Open to inspection.
Five decisions turn independent model judgments into a clear, repeatable ranking with every underlying score available.
- 01Independent shortlists
Each active model names candidates once, using internal knowledge only.
- 02One combined field
Clear duplicates merge. Every unique and valid candidate stays.
- 03Three full-list scores
Each provider scores the complete field in three separate calls.
- 04Median, twice
Each model’s median combines through a second median across the model panel.
- 05Top N, applied last
The requested list length limits official results only after the complete field has been scored.