Image editing point-wise evaluation on ImagenHub
41.84Spearman Rank CorrelationHuman-to-Human
Evaluation Results
| Method | Links | |
|---|---|---|
| Human-to-HumanMethod Category=Upper Bound2025.09 | 41.84 | |
| Gemini-2.5-FlashMethod Category=Proprietary Models2025.09 | 41.62 | |
| GPT-5Method Category=Proprietary Models2025.09 | 40.85 | |
| GPT-4oMethod Category=Proprietary Models2025.09 | 38.21 | |
| EDITREWARDBackbone=Qwen2.5-VL-7B2025.09 | 36.18 | |
| EDITREWARDBackbone=MiMo-VL-7B-SFT2025.09 | 35.2 | |
| ADIEEMethod Category=Open-Source VLMs2025.09 | 34.5 | |
| Qwen2.5-VL-32B-InstMethod Category=Open-Source VLMs2025.09 | 26.87 | |
| Gemini-2.0-FlashMethod Category=Proprietary Models2025.09 | 23.69 | |
| MiMo-VL-7B-SFT-2508Method Category=Open-Source VLMs2025.09 | 22.14 | |
| Qwen2.5-VL-7B-InstMethod Category=Open-Source VLMs2025.09 | 18.59 | |
| ReasonEditZero-shot=true2026.05 | 0.7566 | |
| Gemini-2.5-FlashZero-shot=true2026.05 | 0.4162 | |
| GPT-5Zero-shot=true2026.05 | 0.4085 | |
| GPT-4oZero-shot=true2026.05 | 0.3821 | |
| LMM4EditZero-shot=true2026.05 | 0.3726 | |
| EditReward (MiMo)Zero-shot=true2026.05 | 0.352 | |
| EditScore (Qwen3)Zero-shot=true2026.05 | 0.3062 | |
| Qwen3-VL (7B)Zero-shot=true2026.05 | 0.2426 | |
| Gemini-2.0-FlashZero-shot=true2026.05 | 0.2369 | |
| InternVL3.5 (7B)Zero-shot=true2026.05 | 0.2071 | |
| Qwen2.5-VL-3B-InstMethod Category=Open-Source VLMs2025.09 | -2.54 |