Reasoning-informed Image Editing on RISEBench 1.0 (test)
54.1Temporal Reasoning ScoreGPT-Image-1.5
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| GPT-Image-1.5Model Type=Generation Model2026.03 | 54.1 | 60 | 62 | 21.2 | 50 | 69.7 | 92.5 | 94.9 | |
| Nano Banana ProModel Type=Generation Model2026.03 | 41.2 | 61.1 | 48 | 37.6 | 47.2 | 77 | 85.5 | 94.4 | |
| Seedream 4.0Model Type=Generation Model2026.03 | 12.9 | 12.2 | 11 | 7.1 | 10.8 | 58.9 | 67.4 | 91.2 | |
| Qwen-Image-Edit#Params=20B, Model Type=Generation Model2026.03 | 4.7 | 10 | 17 | 2.4 | 8.9 | 37.2 | 66.4 | 86.9 | |
| InternVL-U#Params=2B+1.7B, Model Type=Unified Model, CoT=true2026.03 | 4.7 | 7.8 | 1.8 | 5.9 | 9.4 | 43.9 | 64.4 | 79.7 | |
| InternVL-U#Params=2B+1.7B, Model Type=Unified Model, CoT=false2026.03 | 3.5 | 2.2 | 5 | 3.5 | 3.6 | 35.6 | 52.7 | 75.9 | |
| Lumina-DiMOO#Params=8B, Model Type=Unified Model2026.03 | 2.4 | 1.1 | 4 | 1.2 | 2.2 | 34 | 50.7 | 72.3 | |
| BAGEL#Params=7B+7B, Model Type=Unified Model2026.03 | 2.4 | 5.6 | 14 | 1.2 | 6.1 | 36.5 | 53.5 | 73 | |
| FLUX.1 Kontext#Params=12B, Model Type=Generation Model2026.03 | 2.3 | 5.5 | 13 | 1.2 | 5.8 | 26 | 71.6 | 85.2 | |
| Ovis-U1#Params=2.4B+1.2B, Model Type=Unified Model2026.03 | 1.2 | 3.3 | 4 | 2.4 | 2.8 | 33.9 | 52.7 | 72.9 |