Instructional Image Editing on RE-Edit
22Physical FidelityEditRefine
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| EditRefineEvaluator=Qwen3-VL-30B, Backbone=FLUX.2 Dev, Executor=Executor-Q (Qwen-Image-Edit)2026.04 | 22 | 14.8 | 8.3 | 16 | 48.5 | 7.7 | 4.4 | |
| Qwen-Image-EditEvaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 21 | 12.8 | 3.3 | 13 | 50 | 7 | 3.8 | |
| EditRefineEvaluator=Qwen3-VL-30B, Backbone=Qwen-Image-Edit, Executor=Executor-F (FLUX.2 Dev)2026.04 | 21 | 13.8 | 6.1 | 17.5 | 45.6 | 7.5 | 4.2 | |
| FLUX.2 DevEvaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 20.1 | 14.8 | 7.8 | 15 | 50.5 | 7.7 | 4 | |
| Nano BananaEvaluator=Qwen3-VL-30B, Category=Commercial Models2026.04 | 20.1 | 10.8 | 3.9 | 15.5 | 48 | 7.6 | 3.6 | |
| EditRefineEvaluator=Qwen3-VL-30B, Backbone=Nano Banana, Executor=Executor-Q (Qwen-Image-Edit)2026.04 | 19.6 | 11.8 | 6.1 | 16.5 | 49.5 | 7.8 | 3.8 | |
| Seedream 4.0Evaluator=Qwen3-VL-30B, Category=Commercial Models2026.04 | 18.7 | 16.3 | 6.1 | 14.5 | 53.4 | 8 | 3.7 | |
| EditRefineEvaluator=Qwen3-VL-30B, Backbone=Seedream 4.0, Executor=Executor-Q (Qwen-Image-Edit)2026.04 | 18.2 | 17.2 | 7.8 | 16.5 | 49.5 | 8 | 4 | |
| Ovis-U1-3BEvaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 16.8 | 13.3 | 2.2 | 4.5 | 39.2 | 7.2 | 4.5 | |
| Step1X-Edit-v1p2-previewEvaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 16.4 | 9.4 | 0.6 | 5 | 47.5 | 7 | 3.9 | |
| Step1X-Edit-v1p1Evaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 15.8 | 7.4 | 2.2 | 7.5 | 38.2 | 6.6 | 3.9 | |
| FLUX.1.KontextEvaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 15.4 | 4.4 | 2.8 | 7.5 | 35.3 | 5.8 | 3.9 | |
| DreamOmni2Evaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 15.4 | 9.4 | 2.2 | 9 | 36.8 | 6 | 3.5 | |
| OmniGen2Evaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 14.5 | 3.4 | 1.1 | 5.5 | 38.2 | 5.4 | 3.6 | |
| HiDream-E1Evaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 14 | 8.9 | 1.1 | 5.5 | 32.4 | 6 | 4.8 | |
| EditRefineEvaluator=Qwen3-VL-30B, Backbone=FLUX.1.Kontext, Executor=Executor-Q (Qwen-Image-Edit)2026.04 | 14 | 6.9 | 4.4 | 10 | 36.8 | 6.6 | 4.3 | |
| Janus-4oEvaluator=Qwen3-VL-30B, Category=Open-source Models2026.04 | 6 | 0.98 | 0 | 0.5 | 7.32 | 3.6 | 4 |