Narrative Product Grid Collage Generation on 500 reference-product pairs curated dataset 1.0 (test)
64.499ArtiMuse ScoreSelf-Reasoning Agentic Framework
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Self-Reasoning Agentic FrameworkBase Model=Seedream 4.52026.04 | 64.499 | 7.321 | 5.886 | 7.584 | 7.371 | 7.255 | 8.362 | 0.932 | |
| Self-Reasoning Agentic FrameworkBase Model=Nano Banana2026.04 | 63.581 | 7.534 | 5.946 | 7.822 | 7.138 | 7.01 | 8.113 | 0.93 | |
| Self-Reasoning Agentic FrameworkBase Model=Flux Kontext Max2026.04 | 62.933 | 7.24 | 5.744 | 7.802 | 6.696 | 6.548 | 7.958 | 0.921 | |
| Self-Reasoning Agentic FrameworkBase Model=Nano Banana Pro2026.04 | 61.16 | 7.529 | 6.103 | 7.813 | 7.462 | 7.171 | 8.203 | 0.934 | |
| Nano Banana2026.04 | 59.184 | 7.101 | 5.328 | 6.788 | 7.119 | 6.624 | 6.273 | 0.88 | |
| Seedream 4.52026.04 | 59.164 | 6.866 | 5.177 | 6.744 | 6.817 | 6.325 | 6.367 | 0.872 | |
| Flux Kontext Max2026.04 | 59.114 | 6.494 | 4.55 | 5.708 | 3.376 | 3.252 | 5.185 | 0.714 | |
| Nano Banana Pro2026.04 | 58.375 | 7.123 | 5.559 | 7.588 | 7.239 | 6.981 | 6.665 | 0.907 |