IDEA-Bench Arena

Arena leaderboard for IDEA-Bench image generation systems, reporting anonymous and full Arena Elo ratings for text-guided image generation/editing pipelines.

7rows
arena_elo_anonymousprimary metric
2026-05-06sampled

Metadata

Metrics

Arena Elo rating (anony), Arena Elo rating (full)

Latest Results

Rank Subject Arena Elo rating (anony) Model Match Provenance Sampled
1 GPT-4o + FLUX.1 [dev] 1058.74 Imported 2026-05-06
2 GPT-4o + Stable Diffusion 3 Medium 1036.95 Imported 2026-05-06
3 GPT-4o + PixArt-Sigma 1036.65 Imported 2026-05-06
4 ChatDiT 1033.22 Imported 2026-05-06
5 GPT-4o + DALLE-3 1024.50 Imported 2026-05-06
6 GPT-4o + Emu2 910.51 Imported 2026-05-06
7 GPT-4o + OmniGen 899.44 Imported 2026-05-06