IDEA-Bench Arena - Anonymous — leaderboard
IDEA-Bench Arena ranks text-to-image generation systems from pairwise human preferences, reporting anonymous and full arena Elo ratings.
Metric: Anonymous Arena Elo. Source: huggingface.co. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o + FLUX.1 [dev] | 1058.7 |
| 2 | GPT-4o + Stable Diffusion 3 Medium | 1037 |
| 3 | GPT-4o + PixArt-Sigma | 1036.6 |
| 4 | ChatDiT | 1033.2 |
| 5 | GPT-4o + DALLE-3 | 1024.5 |
| 6 | GPT-4o + Emu2 | 910.5 |
| 7 | GPT-4o + OmniGen | 899.4 |
Interactive version: theaggregate.ai/benchmark?slug=idea-bench-arena-anonymous · How the rankings work · Data refreshed daily, snapshot 2026-07-22.