Text-to-Image Generation on AtelierEval GI-1 Backend
83.5Objective ScoreHuman
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HumanStrategy=Skilled2026.05 | 83.5 | 4.01 | |
| GPT-5.2Strategy=Skilled2026.05 | 73.4 | 4.22 | |
| Gem-3Strategy=Skilled2026.05 | 72.5 | 4.24 | |
| Cl-4.5Strategy=Skilled2026.05 | 72 | 4.17 | |
| GPT-5.2Strategy=Novice2026.05 | 71.9 | 4.21 | |
| Cl-4.5Strategy=Novice2026.05 | 70.3 | 4.13 | |
| Gem-3Strategy=Novice2026.05 | 69.8 | 4.22 | |
| Qwen-LStrategy=Skilled2026.05 | 69.8 | 4.19 | |
| GPT-4.1Strategy=Skilled2026.05 | 69.4 | 4.25 | |
| Gem-2Strategy=Skilled2026.05 | 68.8 | 4.13 | |
| GPT-4nStrategy=Skilled2026.05 | 68.2 | 4.12 | |
| GPT-4.1Strategy=Novice2026.05 | 67.6 | 4.18 | |
| Qwen-LStrategy=Novice2026.05 | 66.7 | 4.2 | |
| HumanStrategy=Novice2026.05 | 66.4 | 3.25 | |
| Gem-2Strategy=Novice2026.05 | 66.2 | 4.12 | |
| GPT-4nStrategy=Novice2026.05 | 65.7 | 4.17 | |
| Qwen-SStrategy=Novice2026.05 | 64.9 | 4.14 | |
| Qwen-SStrategy=Skilled2026.05 | 64.2 | 4.1 |