Text-to-Image on UniGenBench++ (Short/Long Metrics)
68.85Short Summary MetricDALLE-3
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DALLE-32026.05 | 68.85 | 70.82 | |
| w/ RPO-Gemini 3.1 Pro-ARRBase Generator=FLUX.1-Dev2026.05 | 65.89 | 72.93 | |
| w/ RPO-GPT-5-ARRBase Generator=FLUX.1-Dev2026.05 | 65.36 | 72.41 | |
| w/ RPO-Qwen3vl-8B-ARRBase Generator=FLUX.1-Dev2026.05 | 64.17 | 71.82 | |
| OmniGen22026.05 | 63.09 | 71.39 | |
| Show-o22026.05 | 61.9 | 70.33 | |
| FLUX.1-Dev2026.05 | 60.97 | 69.42 | |
| BAGEL2026.05 | 59.91 | 71.26 | |
| JanusFlow2026.05 | 47.1 | 54.8 | |
| Emu32026.05 | 45.42 | 50.59 |