Text-to-Image Generation on AtelierEval Flux Pro Backend
73.1Objective ScoreHuman
Evaluation results
| Method | Links | ||
|---|---|---|---|
| HumanStrategy=Skilled2026.05 | 73.1 | 3.93 | |
| GPT-5.2Strategy=Novice2026.05 | 64.1 | 3.92 | |
| Gem-3Strategy=Novice2026.05 | 63.8 | 3.98 | |
| GPT-4.1Strategy=Skilled2026.05 | 63.6 | 3.93 | |
| Gem-3Strategy=Skilled2026.05 | 63.2 | 4 | |
| GPT-4.1Strategy=Novice2026.05 | 62.3 | 3.88 | |
| GPT-5.2Strategy=Skilled2026.05 | 62.2 | 3.9 | |
| Cl-4.5Strategy=Skilled2026.05 | 62.1 | 3.84 | |
| Cl-4.5Strategy=Novice2026.05 | 61.1 | 3.83 | |
| Gem-2Strategy=Skilled2026.05 | 60.7 | 3.79 | |
| GPT-4nStrategy=Skilled2026.05 | 60.7 | 3.85 | |
| Qwen-LStrategy=Skilled2026.05 | 60.5 | 3.94 | |
| Qwen-LStrategy=Novice2026.05 | 59.5 | 3.92 | |
| GPT-4nStrategy=Novice2026.05 | 59.3 | 3.78 | |
| Gem-2Strategy=Novice2026.05 | 57.7 | 3.68 | |
| HumanStrategy=Novice2026.05 | 56.8 | 2.94 | |
| Qwen-SStrategy=Novice2026.05 | 54.4 | 3.81 | |
| Qwen-SStrategy=Skilled2026.05 | 54 | 3.82 |