Text-to-Image Generation on Human Preference Evaluation Set benchmark leaderboard · SOTA2 Research

Text-to-Image Generation on Human Preference Evaluation Set

4.39DEQAMaskFocus

Evaluation Results

Method	Links
MaskFocus2025.12		4.39	22.39	35.52	1.09
FLUX.1-dev2025.12		4.37	22.97	31.13	1.06
MaskGRPO2025.12		4.35	22.34	35.48	1.06
SD3.5-M2025.12		4.24	22.5	30.17	0.98
SDXL2025.12		4.17	22.27	30.06	0.56
Meissonic2025.12		4	21.63	28.89	0.39