模型排行

排名	變化	模型	得分	置信區間
1	—	GPT Image 2 (high)閉源OpenAI	1,338	±8
2	▲ 1	MAI-Image-2.5閉源Microsoft AI	1,266	±9
3	▲ 1	HiDream-O1-Image-1.5閉源HiDream	1,264	±9
4	▲ 1	Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)閉源Google	1,259	±10
5	▲ 1	GPT Image 1.5 (high)閉源OpenAI	1,258	±8
6	▲ 1	Nano Banana 2 (Gemini 3.1 Flash Image Preview)閉源Google	1,254	±8
7	▲ 1	Riverflow 2.0閉源Sourceful	1,250	±8
8	▼ 6	Reve 2.0閉源Reve	1,246	±10
9	—	Nano Banana Pro (Gemini 3 Pro Image)閉源Google	1,218	±8
10	▲ 1	Recraft V4.1 Utility Pro閉源Recraft	1,208	±9
11	▼ 1	MAI-Image-2.5-Flash閉源Microsoft AI	1,206	±9
12	—	grok-imagine-image-quality閉源SpaceXAI	1,201	±8
13	▲ 1	Recraft V4.1 Utility閉源Recraft	1,199	±9
14	▼ 1	Krea 2 Medium閉源Krea	1,196	±9
15	—	FLUX.2 [max]閉源Black Forest Labs	1,191	±8
16	▲ 2	Seedream 4.0閉源ByteDance Seed	1,189	±7
17	—	Krea 2 Large閉源Krea	1,187	±8
18	▲ 1	FLUX.2 [pro]閉源Black Forest Labs	1,185	±8
19	▼ 3	MAI-Image-2閉源Microsoft AI	1,185	±8
20	▲ 1	Imagen 4 Ultra Preview 0606閉源Google	1,178	±8
21	▼ 1	FLUX.2 [flex]閉源Black Forest Labs	1,177	±8
22	—	grok-imagine-image閉源SpaceXAI	1,176	±8
23	▲ 1	Qwen Image 2.0 Pro (2026-04-22)閉源Alibaba	1,171	±9
24	▼ 1	Luma UNI 1 Max閉源Luma Labs	1,170	±9
25	—	Imagen 4 Ultra閉源Google	1,170	±8
26	—	ImagineArt 2.0閉源ImagineArt	1,167	±7
27	—	Seedream 4.5閉源ByteDance Seed	1,164	±7
28	—	Imagen 4 Preview 0606閉源Google	1,160	±8
29	—	MAI-Image-2-Efficient閉源Microsoft AI	1,156	±8
30	▲ 1	Nano Banana (Gemini 2.5 Flash Image)閉源Google	1,153	±8
31	▼ 1	Recraft V4.1 Pro閉源Recraft	1,152	±9
32	—	Recraft V4.1閉源Recraft	1,149	±9
33	—	ImagineArt 1.5 Preview閉源ImagineArt	1,147	±8
34	—	Luma UNI 1閉源Luma Labs	1,145	±8
35	—	grok-imagine-image-pro閉源SpaceXAI	1,143	±8
36	▲ 1	Wan 2.5 Preview閉源Alibaba	1,142	±8
37	▼ 1	Seedream 3.0閉源ByteDance Seed	1,141	±8
38	▲ 1	Wan2.6 Text to Image閉源Alibaba	1,133	±8
39	▼ 1	Recraft V4 Pro閉源Recraft	1,133	±8
40	—	Wan 2.6 Image閉源Alibaba	1,132	±8
41	▲ 1	Vivago 2.1閉源HiDream	1,132	±8
42	▼ 1	Recraft V4閉源Recraft	1,132	±8
43	—	GPT Image 1 (high)閉源OpenAI	1,129	±8
44	—	Kolors 2.1閉源KlingAI	1,125	±8
45	—	image-1閉源Api Airforce	1,122	±7
46	—	FLUX.1 Kontext [max]閉源Black Forest Labs	1,121	±8
47	—	Lucid Origin Ultra閉源Leonardo.Ai	1,118	±8
48	—	Seedream 5.0 Lite閉源ByteDance	1,117	±8
49	—	Wan 2.7 Pro閉源Alibaba	1,117	±7
50	—	Lucid Origin Fast閉源Leonardo.Ai	1,109	±8
51	—	Eigen Image閉源Eigen AI	1,106	±8
52	▲ 1	Wan 2.7閉源Alibaba	1,106	±8
53	▲ 1	Vivago 2.0閉源HiDream	1,105	±8
54	▼ 2	Imagen 3 (v002)閉源Google	1,105	±8
55	—	Vidu Q2閉源Vidu	1,101	±8
56	—	Qwen Image 2.0 (2026-03-03)閉源Alibaba	1,099	±9
57	—	Imagen 4 Standard閉源Google	1,097	±8
58	—	Reve Image (Halfmoon)閉源Reve	1,094	±8
59	—	FLUX1.1 [pro] Ultra閉源Black Forest Labs	1,094	±8
60	—	Kling Image 3.0 Omni閉源KlingAI	1,092	±8
61	—	FLUX.1 Kontext [pro]閉源Black Forest Labs	1,089	±8
62	—	GPT Image 1 Mini (medium)閉源OpenAI	1,087	±8
63	▲ 1	FLUX1.1 [pro]閉源Black Forest Labs	1,086	±8
64	▼ 1	Dreamina 3.1閉源ByteDance	1,084	±8
65	—	Imagen 4 Fast Preview 0606閉源Google	1,081	±8
66	—	Qwen Image Plus 2601閉源Alibaba	1,077	±8
67	—	Ideogram 3.0閉源Ideogram	1,075	±8
68	—	Imagen 4 Fast閉源Google	1,071	±8
69	▲ 2	Ideogram v2閉源Ideogram	1,068	±8
70	▼ 1	P-Image閉源Pruna AI	1,067	±8
71	▼ 1	Midjourney v7 Alpha閉源Midjourney	1,066	±8
72	▲ 1	FLUX.1 [pro]閉源Black Forest Labs	1,066	±8
73	▲ 1	Midjourney v6閉源Midjourney	1,064	±8
74	▼ 2	Recraft V3閉源Recraft	1,064	±8
75	—	Luma Photon閉源Luma Labs	1,063	±8
76	—	Ideogram v2 Turbo閉源Ideogram	1,055	±8
77	—	Image-01閉源MiniMax	1,051	±8
78	—	Midjourney v6.1閉源Midjourney	1,049	±8
79	—	MAI Image 1閉源Microsoft AI	1,037	±8
80	—	Z-Image Base閉源Alibaba	1,031	±8
81	—	Ideogram v1閉源Ideogram	1,030	±8
82	—	Phoenix 1.0 Ultra閉源Leonardo.Ai	1,029	±8
83	—	Firefly Image 5 Preview閉源Adobe	1,022	±8
84	—	Stable Diffusion 3 Large閉源Stability.ai	1,021	±8
85	▲ 1	Ideogram v2a閉源Ideogram	1,012	±9
86	▼ 1	Ideogram v2a Turbo閉源Ideogram	1,011	±9
87	—	Playground v3 (beta)閉源Playground AI	1,007	±8
88	—	Phoenix 0.9 Ultra閉源Leonardo.Ai	1,006	±8
89	—	Gemini 2.0 Flash Preview閉源Google	1,005	±12
90	—	Krea 1閉源Krea	1,004	±8
91	—	Firefly Image 4閉源Adobe	1,000	±8
92	—	Phoenix 1.0 Fast閉源Leonardo.Ai	991	±8
93	—	Luma Photon Flash閉源Luma Labs	983	±8
94	▲ 1	Step Image Edit 2閉源StepFun	976	±11
95	▼ 1	Recraft 20B閉源Recraft	975	±8
96	—	Runway Gen-4 Image閉源Runway	975	±9
97	—	DALLE 3 HD閉源OpenAI	957	±8
98	—	Gemini 2.0 Flash Experimental閉源Google	956	±12
99	—	DALLE 3閉源OpenAI	947	±8
100	—	Firefly Image 3閉源Adobe	931	±8
101	—	Grok 2閉源SpaceXAI	921	±14
102	—	Amazon Titan G1 v2 (Standard)閉源Amazon	907	±8
103	—	Stable Diffusion 3 Large Turbo閉源Stability.ai	904	±8
104	—	Stable Diffusion 1.6閉源Stability.ai	900	±9
105	—	Amazon Titan G1 (Standard)閉源Amazon	897	±9
106	—	DALLE 2閉源OpenAI	741	±11

其餘 56 個模型

智力進步

每家廠商當時最強的模型在 Artificial Analysis 智力指數上的分數，按模型釋出的月份畫。

漲得最多 · Anthropic +57.2目前最高 · Anthropic 64.6最高與最低相差 18.1 分時間範圍 2025年11月 – 2026年11月

64.6Anthropic60.0Google56.9OpenAI50.1Meta46.5Kimi