Evaluation of Image Generation Capabilities of Artificial Intelligence Models / 人工智能模型图像生成能力综合评测

by Zhenhui (Jack) Jiang¹, Zhengyu Wu¹, Jiaxin Li¹, Haozhe Xu², Yifan Wu¹,Yi Lu¹/ 蒋镇辉¹，武正昱¹，李佳欣¹，徐昊哲²，吴轶凡¹，鲁艺¹
¹HKU Business School, ²School of Management, Xi'an Jiaotong University
For access to the full research report, please contact Prof. Jiang at jiangz@hku.hk.

Rank	Model	Elo
1	Dreamina (即梦AI)	1123
2	ERNIE Bot (文心一言) V3.2.0	1105
3	Midjourney v6.1	1094
4	DouBao (豆包)	1084
5	MiaoBiShengHua (妙笔生画)	1083
6	FLUX.1 Pro	1079
7	GPT-4o	1058
8	Gemini 1.5 Pro	1045
9	DALL-E3	1025
10	SenseChat (商量) -5	1022
11	SenseMirage (秒画) V5.0	1014
12	Hunyuan-DiT（混元生图）	1005
12	Playground v2.5	1005
14	Imagen 3	1000
15	Stable Diffusion 3 Large	995
16	Spark (讯飞星火)	969
17	CogView3 - Plus	953
17	Qwen (通义千问) V2.5.0	953
19	WenXinYiGe (文心一格2)	890
20	TongYiWanXiang (通义万相)wanx-v2	854
21	360 ZhiHui (360智绘)	834
22	DeepSeek Janus-Pro	810

Rank	Model	Score
1	DouBao (豆包)	5.3
2	Dreamina (即梦AI)	5.2
3	ERNIE Bot (文心一言) V3.2.0	5.16
4	GPT-4o	5.02
5	Gemini 1.5 Pro	4.97
6	MiaoBiShengHua (妙笔生画)	4.71
7	Midjourney v6.1	4.66
7	SenseMirage (秒画) V5.0	4.66
9	CogView3 - Plus	4.58
10	Qwen (通义千问) V2.5.0	4.39
11	TongYiWanXiang (通义万相)wanx-v2	4.25
12	360 ZhiHui(360智绘)	3.85
13	WenXinYiGe (文心一格2)	3.05