Text-to-Image Generation on AtelierEval nBanana Backend
84.9Objective ScoreHuman
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HumanStrategy=Skilled2026.05 | 84.9 | 4.11 | |
| Gem-3Strategy=Skilled2026.05 | 73.9 | 4.15 | |
| GPT-5.2Strategy=Skilled2026.05 | 73.5 | 4.17 | |
| GPT-5.2Strategy=Novice2026.05 | 73.2 | 4.14 | |
| Cl-4.5Strategy=Skilled2026.05 | 72.5 | 4.08 | |
| Qwen-LStrategy=Skilled2026.05 | 70.9 | 4.12 | |
| Gem-3Strategy=Novice2026.05 | 70.6 | 4.1 | |
| GPT-4.1Strategy=Skilled2026.05 | 70.3 | 4.17 | |
| Cl-4.5Strategy=Novice2026.05 | 70 | 4.02 | |
| Gem-2Strategy=Skilled2026.05 | 68.7 | 4.06 | |
| GPT-4nStrategy=Skilled2026.05 | 67.7 | 4.08 | |
| GPT-4.1Strategy=Novice2026.05 | 67.6 | 4.07 | |
| Qwen-LStrategy=Novice2026.05 | 67.1 | 4.13 | |
| Gem-2Strategy=Novice2026.05 | 66 | 3.95 | |
| Qwen-SStrategy=Skilled2026.05 | 65.4 | 4.01 | |
| GPT-4nStrategy=Novice2026.05 | 64.7 | 4.08 | |
| Qwen-SStrategy=Novice2026.05 | 64.1 | 4.05 | |
| HumanStrategy=Novice2026.05 | 58.2 | 3.11 |