Image editing preference evaluation on MM-RewardBench2 Edit
79.2AccuracyGemini 3.1 Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 3.1 ProARR=true2026.05 | 79.2 | |
| GPT-5ARR=true2026.05 | 77.5 | |
| Gemini 3.1 ProARR=false2026.05 | 77.4 | |
| GPT-5ARR=false2026.05 | 73.8 | |
| EditReward2026.05 | 67.2 | |
| Qwen3-VL-8BARR=true2026.05 | 65.5 | |
| Qwen3-VL-8BARR=false2026.05 | 59.2 |