| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| ImgEdit Bench 1.0 (test) | Add Score4.62 | 37 | 4d ago | ||
| InstructPE (test) | DiffChat | PickScore19.836 | 36 | 4d ago | |
| GEditBench English v1 | UniWorld-V2 | G_SC8.29 | 14 | 4d ago | |
| ImgEdit 75 | Z-Image | Add Score4.4 | 13 | 4d ago | |
| EmuEdit-bench (test) | Kontext | CLIP-src Score0.9091 | 13 | 4d ago | |
| GEdit-bench (test) | TAG-MoE | CLIP Score (Source)92.38 | 12 | 4d ago | |
| REDEdit-Bench-CN General dimensions | Overall Score4.48 | 9 | 4d ago | ||
| MagicBrush | MCIE-E1 | L1 Loss0.0641 | 9 | 4d ago | |
| Instruction-based image editing dataset Novel Views averaged across four scenes | SSIM0.8061 | 7 | 4d ago | ||
| Instruction-based image editing dataset averaged across four scenes (train) | SSIM92.15 | 7 | 4d ago | ||
| KRIS-Bench 1.0 (test) | ThinkRL-Edit | Attribute Perception81.02 | 7 | 4d ago | |
| AdvancedEdit-Eval | InsightEdit | VIEScore0.831 | 7 | 4d ago | |
| Human Preference Study Multi-Instruction | FoI | Instruction Align8,083 | 6 | 4d ago | |
| Human Preference Study Single-Instruction | MagicBrush | Instruction Alignment23.17 | 6 | 4d ago | |
| User Study Instruction-based Editing | CogniEdit | Quality3.712 | 4 | 4d ago | |
| FairFace | Qwen-Image-Edit-2511 | Edit Success4.65 | 3 | 4d ago |
