模型排行

排名	變化	模型	得分	置信區間
1	—	GPT Image 2 (high)閉源OpenAI	1,178	±10
2	新	MAI-Image-2.6閉源Microsoft AI	1,149	±12
3	—	Reve 2.1閉源Reve	1,127	±9
4	新	Nano Banana 2 (Gemini 3.1 Flash Image)閉源Google	1,121	±9
5	新	Muse Image閉源Meta	1,116	±11
6	▼ 1	GPT Image 1.5 (high)閉源OpenAI	1,114	±9
7	▼ 1	MAI-Image-2.5閉源Microsoft AI	1,112	±9
8	新	MAI-Image-2.6-Flash閉源Microsoft AI	1,099	±10
9	▼ 2	Nano Banana Pro (Gemini 3 Pro Image)閉源Google	1,099	±9
10	▼ 2	MAI-Image-2.5-Pro閉源Microsoft AI	1,096	±9
11	▼ 2	Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)閉源Google	1,089	±9
12	▼ 2	Qwen-Image-3.0-Pro閉源Alibaba	1,086	±11
13	▼ 2	Seedream 5.0 Pro閉源ByteDance Seed	1,083	±9
14	▼ 1	Qwen-Image-3.0閉源Alibaba	1,078	±11
15	—	grok-imagine-image-quality閉源SpaceXAI	1,034	±8
16	—	Qwen Image 2.0 Pro (2026-04-22)閉源Alibaba	1,032	±10
17	▲ 1	HiDream-O1-Image-1.5閉源HiDream	1,032	±9
18	▼ 1	MAI-Image-2.5-Flash閉源Microsoft AI	1,031	±10
19	—	FLUX.2 [max]閉源Black Forest Labs	1,031	±9
20	—	Seedream 4.0閉源ByteDance Seed	1,029	±11
21	▲ 4	grok-imagine-image閉源SpaceXAI	1,027	±10
22	—	FLUX.2 [flex]閉源Black Forest Labs	1,026	±10
23	—	Krea 2 Large閉源Krea	1,024	±8
24	▼ 3	Luma UNI 1 Max閉源Luma Labs	1,024	±10
25	▲ 4	Krea 2 Medium Turbo閉源Krea	1,023	±10
26	—	Recraft V4.1 Utility閉源Recraft	1,018	±9
27	▼ 3	Recraft V4.1 Utility Pro閉源Recraft	1,018	±8
28	▲ 2	Wan2.6 Text to Image閉源Alibaba	1,012	±10
29	▼ 1	MAI-Image-2閉源Microsoft AI	1,011	±8
30	▲ 1	FLUX.2 [pro]閉源Black Forest Labs	1,011	±9
31	▼ 4	Krea 2 Medium閉源Krea	1,011	±8
32	—	Wan 2.6 Image閉源Alibaba	1,010	±10
33	▲ 1	GPT Image 1 (high)閉源OpenAI	1,007	±10
34	▼ 1	Seedream 4.5閉源ByteDance Seed	1,006	±10
35	▲ 1	P-Image-Ideogram (High)閉源Pruna AI	1,002	±11
36	▲ 1	Seedream 5.0 Lite閉源ByteDance Seed	1,000	±10
37	▼ 2	Luma UNI 1閉源Luma Labs	1,000	±10
38	—	Recraft V4.1閉源Recraft	1,000	±10
39	—	Recraft V4 Pro閉源Recraft	991	±10
40	▲ 1	Nano Banana (Gemini 2.5 Flash Image)閉源Google	991	±10
41	▼ 1	Imagen 4 Ultra閉源Google	990	±8
42	▲ 1	MAI-Image-2-Efficient閉源Microsoft AI	984	±8
43	▼ 1	Recraft V4.1 Pro閉源Recraft	983	±10
44	▲ 1	Wan 2.7 Pro閉源Alibaba	983	±10
45	▲ 1	Recraft V4閉源Recraft	982	±10
46	▼ 2	P-Image-Ideogram (Medium)閉源Pruna AI	974	±11
47	—	ImagineArt 2.0閉源ImagineArt	973	±10
48	▲ 1	Wan 2.7閉源Alibaba	972	±10
49	▲ 1	ImagineArt 1.5 Preview閉源ImagineArt	966	±7
50	▼ 2	P-Image-Ideogram (Low)閉源Pruna AI	962	±11
51	—	Seedream 3.0閉源ByteDance Seed	961	±8
52	—	Vivago 2.1閉源HiDream	952	±8
53	—	Wan 2.5 Preview閉源Alibaba	948	±8
54	▲ 1	Kolors 2.1閉源KlingAI	944	±8
55	▲ 1	image-1閉源Api Airforce	941	±8
56	▼ 2	FLUX.1 Kontext [max]閉源Black Forest Labs	940	±10
57	—	Qwen Image 2.0 (2026-03-03)閉源Alibaba	936	±8
58	—	Vivago 2.0閉源HiDream	927	±9
59	—	Imagen 3 (v002)閉源Google	927	±8
60	▲ 1	Imagen 4 Standard閉源Google	925	±8
61	▼ 1	Lucid Origin Ultra閉源Leonardo.Ai	923	±7
62	—	Eigen Image閉源Eigen AI	922	±8
63	—	Vidu Q2閉源Vidu	916	±7
64	—	Reve Image (Halfmoon)閉源Reve	914	±9
65	▲ 3	P-Image-Ideogram (Very Low)閉源Pruna AI	914	±11
66	▼ 1	Kling Image 3.0 Omni閉源KlingAI	912	±7
67	▼ 1	Lucid Origin Fast閉源Leonardo.Ai	911	±7
68	▲ 2	Dreamina 3.1閉源ByteDance	910	±8
69	▲ 2	Qwen Image Plus 2601閉源Alibaba	910	±8
70	▼ 1	GPT Image 1 Mini (medium)閉源OpenAI	908	±8
71	▲ 1	FLUX1.1 [pro] Ultra閉源Black Forest Labs	907	±8
72	▼ 5	FLUX.1 Kontext [pro]閉源Black Forest Labs	906	±8
73	—	Ideogram 3.0閉源Ideogram	903	±8
74	—	Imagen 4 Fast閉源Google	900	±9
75	▲ 2	Midjourney v7 Alpha閉源Midjourney	894	±9
76	▼ 1	FLUX1.1 [pro]閉源Black Forest Labs	891	±7
77	▼ 1	P-Image閉源Pruna AI	891	±8
78	—	FLUX.1 [pro]閉源Black Forest Labs	884	±8
79	—	Ideogram v2閉源Ideogram	884	±8
80	—	Midjourney v6閉源Midjourney	878	±8
81	—	Ideogram v2 Turbo閉源Ideogram	877	±8
82	—	Recraft V3閉源Recraft	874	±7
83	—	Luma Photon閉源Luma Labs	874	±8
84	—	Image-01閉源MiniMax	872	±9
85	—	Z-Image Base閉源Alibaba	867	±8
86	—	MAI Image 1閉源Microsoft AI	863	±8
87	—	Midjourney v6.1閉源Midjourney	859	±8
88	—	Ideogram v1閉源Ideogram	849	±8
89	—	Stable Diffusion 3 Large閉源Stability.ai	843	±8
90	—	Phoenix 1.0 Ultra閉源Leonardo.Ai	842	±9
91	▲ 1	Ideogram v2a Turbo閉源Ideogram	829	±9
92	▲ 1	Ideogram v2a閉源Ideogram	828	±9
93	▼ 2	Phoenix 0.9 Ultra閉源Leonardo.Ai	825	±8
94	—	Playground v3 (beta)閉源Playground AI	816	±8
95	—	Krea 1閉源Krea	814	±9
96	—	Phoenix 1.0 Fast閉源Leonardo.Ai	809	±8
97	—	Step Image Edit 2閉源StepFun	808	±10
98	—	Luma Photon Flash閉源Luma Labs	803	±8
99	—	Runway Gen-4 Image閉源Runway	793	±9
100	—	Recraft 20B閉源Recraft	787	±8
101	—	DALLE 3閉源OpenAI	769	±8
102	—	DALLE 3 HD閉源OpenAI	768	±8
103	—	Amazon Titan G1 v2 (Standard)閉源Amazon	723	±8
104	—	Stable Diffusion 3 Large Turbo閉源Stability.ai	717	±8
105	▲ 1	Stable Diffusion 1.6閉源Stability.ai	715	±8
106	▼ 1	Amazon Titan G1 (Standard)閉源Amazon	715	±8
107	—	DALLE 2閉源OpenAI	546	±10

其餘 57 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