1
Claude Opus 5.5
59.1
—
—
评测基准榜
Artificial Analysis 公开指数 · 分数越高越好 · 点击可进模型详情 · 最后更新 今天 09:14
名次模型综合智能写代码当助手
2
GPT-6 Astra
56.9
76.9
51
3
GPT-6.1 Sol
54.6
—
—
4
Claude Sonnet 5.5
54.5
—
—
5
Qwen3.8 Max
53.4
68.9
49.9
6
GPT-6.1 Terra
51.3
—
—
7
Gemini 4 Argon
50.4
—
—
8
Claude Fable 5.1
50
80.7
57.2
9
Kimi K3.1 Pro
46.5
—
—
10
Claude Opus 5
44.2
78
56.5
11
GPT-6 Sol
44.2
—
—
12
Muse Wonder
43.6
—
—
13
Ling 3.1 Flash
41.1
—
—
14
Muse Spark 1.3
36.6
75.8
55.5
15
Claude Haiku 5.5
36.5
—
—
16
Grok 4.7
35.5
—
—
17
Tencent Hy4
34.6
—
—
18
Qwen3.8 Max
33.7
76.2
56
19
GPT-5.3 Codex
32.5
—
—
20
GPT-5.6 Terra
32.3
76.7
43.2
21
Claude Opus 4.6
31.9
—
—
22
GLM-5.3
31.7
74.8
53.1
23
Grok 4.6
31
75.9
52.5
24
Gemini 3.8 Flash
30.8
76.3
40.2
25
GPT-5.2
30.4
—
—
26
MiMo-V2.6-Pro
29.9
—
—
27
Claude Opus 4.5
29.1
—
—
28
MiniMax-M3.1-Flash
29.1
—
—
29
DeepSeek V4.1 Flash 1017
28.6
—
—
30
MiMo-V2-Pro
28.6
—
—
31
GPT-5.2 Codex
28.5
—
—
32
Step 5 Preview
28.5
—
—
33
Qwen3.6 Max Preview
28.4
—
—
34
Solar Pro 4
28.2
52.7
—
35
GPT-6 Luna
27.9
—
—
36
GLM-5
27.9
—
—
37
GLM-5.3-Flash
27.8
71.5
50.9
38
MiMo-V2.6-Flash
26.7
—
—
39
GLM-5-Turbo
26.6
—
—
40
Gemini 3 Flash Preview
26.3
—
—
41
Mistral Large 4
26.1
—
—
42
Grok 4.20 0309 v2
25.7
—
—
43
MiMo-V2-Omni-0327
25.1
—
—
44
Muse Spark Open
24.9
—
—
45
GPT-5 Codex
24.9
—
—
46
DeepSeek V4.1 Flash
24.7
—
—
47
Solar Mini 4
24.1
—
—
48
GPT-5.6 Luna
23.8
71.4
42.1
49
GPT-5.1 Codex
23.7
—
—
50
GLM 5V Turbo
23.5
—
—
51
DeepSeek V4 Pro 0813
23.1
68.8
41.3
52
Qwen3.5 27B
22.9
—
—
53
Claude 4.1 Opus
22.8
—
—
54
MiniMax-M2.5
22.8
—
—
55
Grok 4
22.5
—
—
56
Kimi K2 Thinking
22
—
—
57
o3-pro
21.9
—
—
58
Qwen3 Max Thinking
21.3
—
—
59
Qwen3.8 27B
20.1
68.1
45.8
60
MiniMax-M3
18.9
58.6
29.5
61
Nova 2.0 Lite
18.4
23
3.1
62
Beam 1.0
17.7
—
—
63
Kimi K3
17.4
72
29.3
64
Qwen3 Next 80B A3B
16.9
17.4
2.1
65
GPT-5 mini
16.8
15.6
6.8
66
Gemini 2.5 Pro
16.1
33.3
1.6
67
GPT-4.1 mini
14.8
20.2
1.8
68
Nemotron 3.5 Super VL
13.6
—
—
69
Command A+
13.1
27.8
1
70
Inkling Small
12.9
52.9
23.5
71
Nemotron 3 Super 120B A12B
12.8
37.7
1.7
72
Qwen3 235B A22B 2507
12.7
22.1
1.3
73
DeepSeek V4 Pro 0813
12.6
—
—
74
Inkling
12.5
52.1
22.5
75
Claude 3.5 Haiku
12.4
15.9
1.9
76
Mercury 2.5
12.3
—
—
77
Nemotron 3 Ultra 550B A55B
11.8
49.3
20.1
78
Gemini 3.5 Flash-Lite
11.7
49.3
14.3
79
gpt-oss-120b
11.6
30.4
3.7
80
Mercury 2
11.5
31.1
4
81
DeepSeek R1
11.4
24.6
1.1
82
Qwen3.5 9B
11.2
28.7
1.2
83
Ling 3.0 Tiny
11.1
26.5
4.4
84
o3-mini
11
16.3
0.9
85
Trinity Large Thinking
10.8
25.8
1
86
Qwen3 30B A3B 2507
9.8
12.1
0.9
87
DeepSeek V3 0324
9.7
21.2
0.8
88
GPT-4.1 nano
9.6
11.1
1.2
89
Llama 3.3 Instruct 70B
9.4
11.9
0.3
90
Llama 4 Maverick
9.3
16.3
0.6
91
Mistral Large 3
9.3
20.1
1
92
Mistral Medium 3.1
9.2
20.5
1.9
93
Qwen3 Coder Next
9.2
36.2
0.9
94
gpt-oss-20b
9
20.7
1.2
95
NVIDIA Nemotron 3 Nano 30B A3B
8.9
14.4
1
96
Devstral 2
8.6
31.3
2.3
97
DeepSeek V3
8.5
23
0.8
98
Solar Pro 3
7.8
16.2
1.4
99
Llama 3.1 Instruct 8B
7.6
5.4
0.5
100
Devstral Small 2
7.5
29.3
2
AI++指数
全部前沿
全部最新模型
行业动态
全部- 产业观察 · 2026-09-13用量榜上的「可控 AI 编程」：CodeGPT 类多模型 + 私有部署编程助手，正在改写 IDE 插件之争
- 产业观察 · 2026-09-09个人 Agent 网关成新品类：OpenClaw 养虾热潮与自托管智能体的入口之争
- 竞争格局 · 2026-09-08开源 Coding Agent 2026：Qwen Code 入场，Claude Code 的护城河被三条路同时撬动
- 产业观察 · 2026-09-01AI 3D 生成迎来产业化分水岭：VAST 30 亿融资 + Tripo P2.0，影眸同日发布世界生成模型
- 产业观察 · 2026-09-05Bun 1.4 把运行时变成『全栈平台』：JS 运行时之战从『跑得快』卷到『什么都不用装』