Image Editing Quality Assessment on GenAI-Bench
83.9AccuracyReasonEdit
Evaluation Results
| Method | Links | |
|---|---|---|
| ReasonEditZero-shot=true2026.05 | 83.9 | |
| EditReward (MiMo)Zero-shot=true2026.05 | 65.72 | |
| LMM4EditZero-shot=true2026.05 | 63.27 | |
| GPT-5Zero-shot=true2026.05 | 59.61 | |
| Gemini-2.5-FlashZero-shot=true2026.05 | 57.01 | |
| GPT-4oZero-shot=true2026.05 | 53.54 | |
| Gemini-2.0-FlashZero-shot=true2026.05 | 53.32 | |
| EditScore (Qwen3)Zero-shot=true2026.05 | 50.24 | |
| InternVL3.5 (7B)Zero-shot=true2026.05 | 45.17 | |
| Qwen3-VL (7B)Zero-shot=true2026.05 | 44.58 |