Image-Augmented Generation on AtelierEval Image Tasks
76.7Objective ScoreHuman
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HumanStrategy=Skilled2026.05 | 76.7 | 3.97 | |
| GPT-5.2Strategy=Novice2026.05 | 65.7 | 3.96 | |
| GPT-5.2Strategy=Skilled2026.05 | 65.4 | 3.96 | |
| Gem-3Strategy=Skilled2026.05 | 65.4 | 4 | |
| Cl-4.5Strategy=Skilled2026.05 | 65.1 | 3.92 | |
| Gem-3Strategy=Novice2026.05 | 64.7 | 4.01 | |
| Cl-4.5Strategy=Novice2026.05 | 64.1 | 3.88 | |
| GPT-4.1Strategy=Skilled2026.05 | 63.9 | 3.99 | |
| Qwen-LStrategy=Skilled2026.05 | 63.1 | 3.98 | |
| GPT-4.1Strategy=Novice2026.05 | 63 | 3.92 | |
| Gem-2Strategy=Skilled2026.05 | 62.9 | 3.89 | |
| GPT-4nStrategy=Skilled2026.05 | 62.8 | 3.92 | |
| Qwen-LStrategy=Novice2026.05 | 61.3 | 3.99 | |
| Gem-2Strategy=Novice2026.05 | 60.9 | 3.82 | |
| GPT-4nStrategy=Novice2026.05 | 60.8 | 3.89 | |
| Qwen-SStrategy=Novice2026.05 | 57.8 | 3.9 | |
| Qwen-SStrategy=Skilled2026.05 | 57.2 | 3.87 | |
| HumanStrategy=Novice2026.05 | 56.9 | 3.01 |