Polymind Back to app
LEADERBOARD

Which model wins most often?

Every Polymind run ends with a judge model picking the panelist(s) it leaned on. We tally those picks across every real run from the last 30 days. Win rate is judge picks divided by appearances as a panelist — not a vote, a synthesis preference.

#ModelPicksRunsWin rate
  1. 1Claude71258.3%
  2. 2Gemini51241.7%
  3. 3GPT41233.3%
  4. 4Grok31030.0%
  5. 5Perplexity31030.0%
  6. 6Mistral1128.3%

13 completed runs in the last 30 days · refreshed every 5 minutes

How we count

Appearances — number of completed runs where this provider was one of the panelists.

Picks — number of those runs where the judge model named this provider as one of the panelists it leaned on most (up to three per run).

Win rate— picks ÷ appearances. A model that's never picked is at the bottom; a model that gets picked every time it appears tops the board. Sample sizes vary; trust the order, not the decimals.