Text-to-Image Factuality Evaluation on ABO 50
88.23FAGER ScoreFLUX1-dev + FAGER
Evaluation Results
| Method | Links | |
|---|---|---|
| FLUX1-dev + FAGERModel=FLUX1-dev, Refinement Backbone=Kontext2026.05 | 88.23 | |
| FLUX1-dev + FAGERModel=FLUX1-dev, Refinement Backbone=Qwen-Edit2026.05 | 85.74 | |
| FLUX2-devModel=FLUX2-dev2026.05 | 85.17 | |
| Nano Banana ProModel=Nano Banana Pro2026.05 | 82.26 | |
| Stable Diffusion 3.5 LargeModel=Stable Diffusion 3.5 Large2026.05 | 65.2 | |
| FLUX1-devModel=FLUX1-dev2026.05 | 62.16 |