M
ModelSage
排行榜/Arena 人类盲测/Code 编程/HTML

Arena Code 编程 · HTML

前端 / 全栈代码生成盲测，含 11 个细分场景子榜 · 完整榜单 142 个模型

Overall 总榜143Fullstack 全栈82Frontend 前端143HTML142React127品牌营销站142参考图还原142数据分析应用142消费级产品142游戏142仿真模拟142内容创作工具138
RankModelElo 分
1
gpt-6.1-sol-max1795.3±34.7
2
gpt-6-astra-max1780.1±22.9
3
claude-opus-5.5-max1763±64
4
claude-sonnet-5.5-xhigh1737.7±36.6
5
muse-wonder-max1721.9±51.5
6
claude-sonnet-5.5-high1710±25.6
7
claude-fable-5.1-max1690.4±23.8
8
gpt-6-sol-max1682.9±24.8
9
claude-opus-5-max1679.8±14.9
10
muse-spark-1.3-max1657.1±19.6
11
claude-opus-5-high1652.1±12.3
12
qwen3.8-max-09021651.3±14.7
13
claude-fable-5-high1650.2±14.9
14
kimi-k3-max1645.8±14
15
mimo-v2.6-flash1643.4±24.4
16
gemini-4-argon-high1639.3±48
17
grok-4.7-xhigh1628.1±24.4
18
muse-spark-1.3 (xHigh)1627.4±19
19
qwen3.8-max1626.7±33.5
20
hy4-preview1621.7±24.6
21
mimo-v2.6-pro1621±25.6
22
gemini-3.8-flash-high1620.1±17.6
23
gpt-5.6-sol-xhigh (codex-harness)1616.6±13.8
24
deepseek-v4.1-flash-max1611.8±26.2
25
qwen3.8-flash-next1603.5±18.4
26
grok-4.6-high1598.8±16.5
27
glm-5.3-max1595.5±18.6
28
gpt-6-luna-max1594.4±13.3
29
gemini-3.7-flash-high1594.3±15.9
30
glm-5.3-flash1578.1±13.7
31
qwen3.8-27b1571.1±13.8
32
grok-4.51567.1±16.1
33
claude-haiku-5.5-high1566.5±32.5
34
glm-5.2-max1566±14.9
35
deepseek-v4-pro-high-202608131564.7±26.9
36
gemini-3.5-flash1559±37
37
claude-opus-4-6-high1558.7±13.2
38
deepseek-v4-flash-high1558.2±25.7
39
claude-opus-4-71554.3±13.4
40
gemini-3.6-flash-high1548.1±20.9
41
claude-opus-4-61546.2±12.5
42
step-5-preview-high1546±25.6
43
claude-opus-4-7-high1545.6±13.5
44
gpt-5.5-xhigh (codex-harness)1543.6±14.9
45
claude-opus-4-8-high1543.2±13.8
46
muse-spark-1.2 (xHigh)1542.9±41.7
47
muse-spark-1.11540.4±21.4
48
gpt-5.5-high (codex-harness)1533.1±14.3
49
claude-sonnet-5-high1528.1±14.5
50
gpt-5.6-terra-xhigh (codex-harness)1527.7±15.1
51
claude-sonnet-4-61526.1±11.9
52
claude-opus-4-81524.6±12.8
53
gemini-3.5-flash-medium1523.1±17.1
54
gpt-5.6-luna-xhigh (codex-harness)1519.9±15.4
55
gemini-3.5-flash-high1513.5±19
56
qwen3.7-max-202605171510.7±19.2
57
claude-opus-4-5-20251101-high-32k1510±12.5
58
glm-5.11509.1±18.2
59
gpt-5.5 (codex-harness)1492.9±14.3
60
qwen3.6-max-preview1492.2±33.8
61
seed-2.1-pro-preview1491.2±14.4
62
mistral-large-41489.4±41.3
63
gemini-3.1-pro-preview1488.9±12.3
64
kimi-k2.61484.4±18.2
65
hy31484.4±20.9
66
claude-opus-4-5-202511011482.6±11.7
67
gpt-5.4-medium (codex-harness)1477.8±52.9
68
gpt-5.4-high (codex-harness)1474.1±59.3
69
mimo-v2.5-pro1472.3±12.1
70
gemini-3-pro1471.8±11.6
71
deepseek-v4-pro-high-preview1470.1±15.8
72
qwen3.6-plus1463.6±13.6
73
minimax-m31455.6±12.9
74
kimi-k2.7-code1450.8±24.1
75
glm-51450.3±20.7
76
glm-4.71446.5±13.5
77
gemini-3-flash1445.3±11.8
78
mimo-v2-pro1439.8±21.1
79
gpt-5.41434.8±27.3
80
deepseek-v4-pro1434.1±15.9
81
mimo-v2.51430±15
82
kimi-k2.5-thinking1427±12.5
83
gpt-5.21426.7±23.5
84
gpt-5-medium1426.5±16.7
85
gpt-5.3-codex (codex-harness)1421.6±31.3
86
gpt-5.4-mini-high1419.9±16.6
87
gpt-5.3-codex (codex-harness)1418.2±36.9
88
deepseek-v4-flash-high-preview1416.2±18
89
minimax-m2.51413.5±20.3
90
kimi-k2.5-instant1410.6±28.5
91
qwen3.5-397b-a17b1409.6±12
92
gpt-5.1-medium1404.9±13
93
minimax-m2.71401.2±14
94
inkling1399.7±14.8
95
claude-sonnet-4-5-20250929-high-32k1398.9±11.1
96
minimax-m2.1-preview1398.8±12.7
97
claude-opus-4-1-202508051398.6±12.2
98
gemini-3-flash (thinking-minimal)1396.2±9.9
99
Inkling Small1395.9±25.7
100
claude-sonnet-4-5-202509291389.5±10.8
101
qwen3.5-27b1386.7±20.4
102
gemma-4-31b1376.1±15
103
deepseek-v3.2-thinking1370.5±13.9
104
solar-pro41369.4±23.5
105
grok-4.20-beta-0309-reasoning1368.5±15.1
106
grok-4.31366.8±15.4
107
gpt-5.11363.7±11.4
108
muse-glimmer1363.2±45
109
gemma-4-26b-a4b1361.9±43.8
110
hunyuan-hy3-preview1354±43.9
111
glm-5v-turbo1354±42.9
112
mimo-v2-flash (non-thinking)1352.4±13.5
113
qwen3.5-122b-a10b1352.2±20.1
114
glm-4.61350.2±12.4
115
gpt-5.2-codex1348.3±15.2
116
mimo-v2-flash (thinking)1346.4±23.3
117
gpt-5.1-codex1346±13
118
laguna-m.11334.7±25.3
119
kimi-k2-thinking-turbo1324.9±11.2
120
claude-haiku-4-5-202510011317.6±9.2
121
mistral-medium-3.51307.4±41.4
122
minimax-m21306.4±12.4
123
deepseek-v3.21305.8±12.4
124
qwen3.5-35b-a3b1301.3±44.8
125
qwen3.5-flash1290.7±59
126
qwen3-coder-480b-a35b-instruct1286.9±11.1
127
laguna-xs.21283.2±29.1
128
deepseek-v3.2-exp1281.2±14.8
129
gemini-3.1-flash-lite-preview1276.5±16.3
130
KAT-Coder-Pro-V11265±20.4
131
grok-4-1-fast-reasoning1254.1±13.9
132
gpt-5.1-codex-mini1252.7±22.8
133
trinity-large-thinking1240.2±49.1
134
mistral-large-31238.4±26.1
135
gemini-2.5-pro1234.7±17
136
grok-4.1-thinking1224.3±25.6
137
devstral-21218.2±23.3
138
granite-4.1-8b1205.1±48.6
139
mercury-21199.1±66.5
140
grok-code-fast-11173.3±28.8
141
grok-4-fast-reasoning1166.6±28.1
142
devstral-medium-25071077.7±31.1
Elo 分 真人在不知道模型名字的情况下两两投票，胜负累积出的等级分。绝对值没有意义，差值才有：分差 100 分 ≈ 强的那个有 64% 胜率，只差几分基本是平手。
± 后面的数 是 95% 置信区间的半宽，投票越多越小。两个模型的区间重叠，说明它们的名次差别在统计上不成立。
区间 把上面这个置信区间换算成名次：1–5 意思是这个模型的真实名次可能在第 1 到第 5 之间。
数据更新: 2026-10-14数据来源: Arena AI ↗