Text-to-Image Generation on Taxonomy-based Evaluation Suite 43 prompts benchmark leaderboard · SOTA2 Research

Text-to-Image Generation on Taxonomy-based Evaluation Suite 43 prompts

0.992Entity AdherenceWan-2.5-preview

Evaluation Results

Method	Links
Wan-2.5-preview2026.05		0.992	0.92	0.793	0.96	0.976	0.875	1	1	0.778	1	0.743	0.969	0.88	0.783	0.615	0.98	0.997	0.85	0.895
Nano-Banana2026.05		0.979	0.9	0.773	0.978	0.95	0.955	1	0.938	0.688	1	1	0.967	0.901	0.875	0.639	0.93	0.998	0.879	0.908
FLUX.2-dev2026.05		0.934	0.936	0.831	0.88	0.743	0.907	0.933	0.688	0.882	0.867	0.848	0.933	0.908	0.817	0.659	0.941	0.996	0.824	0.863
Z-Image-Turbo2026.05		0.913	0.827	0.73	0.705	0.561	0.625	0.935	0.875	0.905	1	0.176	0.767	0.835	0.691	0.485	0.877	0.995	0.796	0.761