Text-to-Image Generation on AtelierEval SDXL Backend
65.4Objective ScoreHuman
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HumanStrategy=Skilled2026.05 | 65.4 | 3.84 | |
| Cl-4.5Strategy=Novice2026.05 | 54.8 | 3.56 | |
| GPT-4.1Strategy=Novice2026.05 | 54.6 | 3.58 | |
| GPT-4nStrategy=Skilled2026.05 | 54.6 | 3.6 | |
| Gem-3Strategy=Novice2026.05 | 54.5 | 3.75 | |
| Gem-2Strategy=Novice2026.05 | 53.8 | 3.55 | |
| Cl-4.5Strategy=Skilled2026.05 | 53.8 | 3.6 | |
| GPT-5.2Strategy=Novice2026.05 | 53.6 | 3.56 | |
| GPT-4nStrategy=Novice2026.05 | 53.6 | 3.53 | |
| Gem-2Strategy=Skilled2026.05 | 53.2 | 3.56 | |
| GPT-5.2Strategy=Skilled2026.05 | 52.6 | 3.54 | |
| GPT-4.1Strategy=Skilled2026.05 | 52.5 | 3.62 | |
| Gem-3Strategy=Skilled2026.05 | 52.1 | 3.62 | |
| Qwen-LStrategy=Novice2026.05 | 51.8 | 3.71 | |
| Qwen-LStrategy=Skilled2026.05 | 50.9 | 3.69 | |
| Qwen-SStrategy=Novice2026.05 | 48.1 | 3.6 | |
| HumanStrategy=Novice2026.05 | 46.3 | 2.74 | |
| Qwen-SStrategy=Skilled2026.05 | 45.4 | 3.57 |