According to AIatMeta, WildArtifactBench uses win rates and Elo from human and agent judges to benchmark multimodal agents across 10 real tasks.

According to emollick, Kimi K3 leads Arena frontend ELO, but user-judged scores are limited for benchmarking and can favor prompt-tuned chat UX.

According to emollick, GDPval AA v2 relies on AI judges and unclear human ELO, limiting trust in Intelligence Index rankings.