Contextual Visual Personalization (Editing) on VPTT-Bench 1.0 (test)
0.626VPTTscore (V)VPRAG
Evaluation results
| Method | Links | ||
|---|---|---|---|
| VPRAGModel=Qwen2026.01 | 0.626 | — | |
| VPRAGModel=4o-mini2026.01 | 0.626 | — | |
| VPRAGModel=Qwen2026.01 | 0.626 | — | |
| VPRAGModel=4o-mini2026.01 | 0.626 | — | |
| Comb. (BRAG + VPRAG)Model=4o-mini2026.01 | 0.61 | 0.5 | |
| Comb. (BRAG + VPRAG)Model=4o-mini2026.01 | 0.61 | 0.5 | |
| Comb. (BRAG + VPRAG)Model=Gemini2026.01 | 0.606 | — | |
| Comb. (BRAG + VPRAG)Model=Gemini2026.01 | 0.606 | — | |
| VPRAGModel=Gemini2026.01 | 0.605 | 0 | |
| VPRAGModel=Gemini2026.01 | 0.605 | 0 | |
| BRAGModel=4o-mini2026.01 | 0.596 | 0.9 | |
| BRAGModel=4o-mini2026.01 | 0.596 | 0.9 | |
| Comb. (BRAG + VPRAG)Model=Qwen2026.01 | 0.586 | 1 | |
| Comb. (BRAG + VPRAG)Model=Qwen2026.01 | 0.586 | 1 | |
| BRAGModel=Qwen2026.01 | 0.583 | 1.1 | |
| BRAGModel=Gemini2026.01 | 0.583 | 0.6 | |
| BRAGModel=Qwen2026.01 | 0.583 | 1.1 | |
| BRAGModel=Gemini2026.01 | 0.583 | 0.6 | |
| Persona OnlyModel=4o-mini2026.01 | 0.384 | 8.8 | |
| Persona OnlyModel=4o-mini2026.01 | 0.384 | 8.8 | |
| Persona OnlyModel=Qwen2026.01 | 0.378 | 8.7 | |
| Persona OnlyModel=Qwen2026.01 | 0.378 | 8.7 | |
| Persona OnlyModel=Gemini2026.01 | 0.372 | 8.1 | |
| Persona OnlyModel=Gemini2026.01 | 0.372 | 8.1 | |
| BaselineModel=Qwen2026.01 | 0.306 | 12 | |
| BaselineModel=4o-mini2026.01 | 0.306 | 12 | |
| BaselineModel=Gemini2026.01 | 0.306 | 10.7 | |
| BaselineModel=Qwen2026.01 | 0.306 | 12 | |
| BaselineModel=4o-mini2026.01 | 0.306 | 12 | |
| BaselineModel=Gemini2026.01 | 0.306 | 10.7 |