Leaderboard
Aggregate ELO per model across every blind matchup. Prompts stay hidden while their arena is live.
Rounds
| Round | Status | Tracks | Votes | Prompt |
|---|
Prompts stay sealed while a round is live and are revealed forever once it retires.
Standings
Bradley-Terry fit over all matchups · 95% CI · matchups never cross rounds| Rank | Model | Rating | Matches | Win rate |
|---|
ELO over time
Win rate
Head-to-head wins
Read across: how many matchups the row model won against the column model.