everyais

arena-vision

Vision Arena

Pairwise human preference for multimodal responses.

Methodology

Scores are shown exactly as published by each source. Source boards are ranked independently and are never averaged.

Independent source results

LMArena Leaderboard

Data source: LMArena LeaderboardLicense: CC-BY-4.0

Synced by EveryAI: This is the sync time, not the measurement date.

LMArena Leaderboard leaderboard
#ModelScore
1Claude Fable 5everyais/claude-fable-51,329 Elo
2GLM-5.3-Flasheveryais/glm-5-3-flash1,296 Elo
3Claude Opus 4.8everyais/claude-opus-4-81,289 Elo
Vision Arena | everyais