🥇 UniREditBench Leaderboard
📚 UniREditBench is a unified benchmark for reasoning-base image editing that integrates diverse prompt themes with a comprehensive suite of dual-reference evaluation pipeline.
🔧 You can use the official GitHub repo to evaluate your model on UniREditBench.
📝 To add your own model to the leaderboard, please send an Email to Feng Han and Yibin Wang, then we will help with the evaluation and updating the leaderboard.
2025-11 | ✓ | 78.15 | 81.08 | 72.39 | 84.78 | 70.81 | 73.85 | 72.73 | 83.13 | 71.27 | 71.62 | 75.74 | 84.90 | 87.77 | 57.68 | 73.05 | 86.80 | 71.16 | 93.20 | 98.30 | 71.47 | 80.48 |
📙 Citation
If you use UniREditBench in your research, please cite our work: