Text-to-Image Generation on DialectGen Concise Prompts benchmark leaderboard · SOTA2 Research

Text-to-Image Generation on DialectGen Concise Prompts

38.63Human ScoreDALL-E 2

Evaluation Results

Method
DALL-E 22025.10	38.63	32.79	9.97	35.87	7.95	48.78	47.21	24.14
Flux.1 [dev]2025.10	36.43	32.26	10.88	30.61	10.83	44.64	42.59	32.62
DALL-E Mini2025.10	34.29	31.52	11.71	33.91	8.18	47.11	42.85	25.51
Stable Diffusion 3.5 Large Turbo2025.10	32.92	30.28	11.34	30.33	9.27	43.6	42.49	25.72
Stable Diffusion 3.5 Large2025.10	32.31	29.43	11.37	28.3	9.74	42.66	41.9	24.56
Stable Diffusion 32025.10	31.89	29.01	10.81	27.89	8.64	42.67	40.69	25.12
Stable Diffusion 2.12025.10	31.46	28.79	11.7	24.35	9.31	44.82	41.12	28.89
Stable Diffusion XL2025.10	29.8	26.69	10.88	23.37	7.95	41.22	38.74	22.17
Stable Diffusion 1.52025.10	29.77	27.06	10.32	19.51	8.66	36.5	42.15	28.48
Stable Diffusion 1.42025.10	28.19	26.7	10.35	20.67	9.64	34.94	41.27	26.96
DALL-E 32025.10	26.55	24.39	9.32	18.97	3.58	41.95	31.9	25.56
gpt-image-1 (4o Image Gen)2025.10	22.18	19.18	7.65	26.12	5.2	26.09	26.51	11.99
DALL-E 32025.10	20.19	18.25	6.69	22.11	6.48	26.86	23.05	12.74