排行榜/Arena 人类盲测/Code 编程/Fullstack 全栈

Arena Code 编程 · Fullstack 全栈

前端 / 全栈代码生成盲测，含 11 个细分场景子榜 · 完整榜单 82 个模型

Overall 总榜143 Fullstack 全栈82 Frontend 前端143 HTML142 React127 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

Rank	区间	Model	厂商	License	Elo 分	投票数	输入 $/M	输出 $/M	上下文
1	1–4	gpt-6.1-sol-max	OpenAI	Proprietary	1759.8±28.8	484	$2	$10	1M
2	1–4	gpt-6-astra-max	OpenAI	Proprietary	1750.4±19.5	1.1k	$10	$50	1M
3	1–5	claude-opus-5.5-max	Anthropic	Proprietary	1742.6±24.8	651	$4	$20	1M
4	1–8	gpt-6-sol-max	OpenAI	Proprietary	1711±21.6	838	$2	$10	1M
5	3–10	claude-sonnet-5.5-xhigh	Anthropic	Proprietary	1697.7±25.6	595	$2	$10	1M
6	4–13	qwen3.8-max	Alibaba	Proprietary	1688.1±29.1	503	$2	$6	1M
7	4–13	gemini-4-argon-high	Google	Proprietary	1685.6±28	480	$2	$10	1M
8	4–13	claude-fable-5.1-max	Anthropic	Proprietary	1680.9±16	1.5k	$10	$50	1M
9	5–13	qwen3.8-max-0902	Alibaba	Proprietary	1672.8±12.6	2.5k	$2	$6	-
10	5–13	claude-opus-5-max	Anthropic	Proprietary	1666.8±10.4	3.9k	$5	$25	1M
11	6–14	kimi-k3-max	Moonshot	Kimi K3 license	1656.7±10.8	3.5k	$3	$15	1M
12	6–21	muse-spark-1.3-max	Meta	Proprietary	1654.2±14.6	1.8k	$1.3	$4.3	1M
13	6–24	claude-sonnet-5.5-high	Anthropic	Proprietary	1651.3±20.1	887	$2	$10	1M
14	12–27	claude-opus-5-high	Anthropic	Proprietary	1635.3±9.6	4.5k	$5	$25	1M
15	11–32	muse-spark-1.3 (xHigh)	Meta	Proprietary	1631.8±14.9	1.6k	$1.3	$4.3	1M
16	12–30	gpt-5.6-sol-xhigh (codex-harness)	OpenAI	Proprietary	1630.9±10.4	3.6k	$4	$20	1M
17	12–32	gemini-3.7-flash-high	Google	Proprietary	1627.1±13	2.3k	$0.75	$3.8	1M
18	12–34	grok-4.7-xhigh	SpaceXAI	Proprietary	1626±19	1.0k	$2	$6	500K
19	12–33	glm-5.3-max	Z.ai	MIT	1625.9±13.8	1.9k	$1.4	$4.4	1M
20	12–34	deepseek-v4.1-flash-max	DeepSeek	MIT	1623.3±21.9	719	$0.30	$1.2	1M
21	12–34	mimo-v2.6-pro	Xiaomi	MIT	1622±19.8	918	$0.43	$0.87	1M
22	13–34	mimo-v2.6-flash	Xiaomi	MIT	1620.2±19.3	965	$0.14	$0.28	1M
23	13–34	qwen3.8-flash-next	Alibaba	qwen-community-1.0	1618.2±15.5	1.5k	$0.16	$0.47	1M
24	14–34	grok-4.6-high	SpaceXAI	Proprietary	1616.1±14.1	1.8k	$2	$6	500K
25	14–34	claude-fable-5-high	Anthropic	Proprietary	1614.9±12.4	2.4k	$10	$50	1M
26	14–34	qwen3.8-27b	Alibaba	Apache 2.0	1614.6±11.4	3.0k	$0.40	$3	-
27	15–34	glm-5.2-max	Z.ai	MIT	1613.7±11.5	3.0k	$1.4	$4.4	1M
28	13–36	deepseek-v4-flash-high	DeepSeek	MIT	1612.9±19.8	918	$0.30	$1.2	1M
29	15–34	gemini-3.8-flash-high	Google	Proprietary	1611±13.3	2.1k	$0.75	$3.8	1M
30	16–35	gpt-6-luna-max	OpenAI	Proprietary	1607±10.4	3.9k	$0.10	$0.50	1M
31	15–42	hy4-preview	Tencent	Apache 2.0	1604.1±19.3	942	$0.83	$2.5	1M
32	18–36	glm-5.3-flash	Z.ai	MIT	1603.3±10.8	3.3k	$0.06	$0.20	1M
33	16–44	deepseek-v4-pro-high-20260813	DeepSeek	MIT	1598.8±18.8	1.0k	$1.3	$4.0	-
34	19–45	muse-spark-1.1	Meta	Proprietary	1593.8±15.1	1.6k	$1.3	$4.3	-
35	30–50	gemini-3.6-flash-high	Google	Proprietary	1582.6±13.1	2.2k	$0.75	$3.8	1M
36	32–50	claude-opus-4-8-high	Anthropic	Proprietary	1580±10.7	4.1k	$5	$25	1M
37	32–52	claude-opus-4-6-high	Anthropic	Proprietary	1576.5±15	1.6k	$5	$25	1M
38	32–52	claude-sonnet-4-6	Anthropic	Proprietary	1576±14.9	1.6k	$1.5	$7.5	1M
39	32–55	claude-opus-4-6	Anthropic	Proprietary	1571.5±15.1	1.6k	$5	$25	1M
40	29–57	mistral-large-4	Mistral	Proprietary	1571±26.4	497	$0.68	$2.1	1M
41	33–55	claude-opus-4-7	Anthropic	Proprietary	1570.5±14.1	1.8k	$5	$25	1M
42	33–55	claude-opus-4-7-high	Anthropic	Proprietary	1570.1±14.7	1.7k	$5	$25	1M
43	32–58	claude-haiku-5.5-high	Anthropic	Proprietary	1566.5±23.3	647	$0.10	$0.50	1M
44	32–57	step-5-preview-high	StepFun	Proprietary	1566.2±21.1	803	$1	$2.7	1M
45	35–55	seed-2.1-pro-preview	Bytedance	Proprietary	1565.8±12.8	2.2k	$0.88	$4.4	262K
46	35–55	gpt-5.6-terra-xhigh (codex-harness)	OpenAI	Proprietary	1565.4±11.6	2.8k	$2	$12	1M
47	35–55	claude-sonnet-5-high	Anthropic	Proprietary	1565.1±11.9	2.6k	$2	$10	1M
48	35–55	gpt-5.6-luna-xhigh (codex-harness)	OpenAI	Proprietary	1564.4±11.4	2.8k	$0.20	$1.2	1M
49	34–58	kimi-k2.6	Moonshot	Modified MIT	1561.1±17.8	1.6k	$0.95	$4	262K
50	35–57	grok-4.5	SpaceXAI	Proprietary	1558.7±13.9	1.8k	$2	$6	500K
51	37–57	claude-opus-4-8	Anthropic	Proprietary	1557.1±12	2.4k	$5	$25	1M
52	37–62	glm-5.1	Z.ai	MIT	1548.3±17.1	1.2k	$1.4	$4.4	203K
53	39–61	mimo-v2.5-pro	Xiaomi	MIT	1546±11.5	2.8k	$0.43	$0.87	1M
54	39–62	hy3	Tencent	Apache 2.0	1542.2±16.9	1.2k	$0.13	$0.53	262K
55	39–63	qwen3.7-max-20260517	Alibaba	Proprietary	1540.4±18.2	1.1k	$1.5	$4.4	1M
56	46–62	minimax-m3	MiniMax	MiniMax Community License	1539.5±11.5	2.8k	$0.30	$1.2	524K
57	46–64	mimo-v2.5	Xiaomi	MIT	1534.4±15.4	1.5k	$0.14	$0.28	1M
58	50–67	gemini-3.5-flash-high	Google	Proprietary	1527.1±16.4	1.4k	$0.75	$4.5	1M
59	52–67	gpt-5.5-xhigh (codex-harness)	OpenAI	Proprietary	1527±15.1	1.5k	$5	$30	-
60	52–67	gemini-3.5-flash-medium	Google	Proprietary	1526.8±14	2.1k	$0.75	$4.5	1M
61	53–68	deepseek-v4-pro-high-preview	DeepSeek	MIT	1516±18	1.1k	$1.3	$4.0	1M
62	52–73	gemini-3.5-flash	Google	Proprietary	1510.8±29.3	407	$0.75	$4.5	1M
63	57–70	gpt-5.5-high (codex-harness)	OpenAI	Proprietary	1508.2±13.2	3.2k	$5	$30	-
64	58–70	qwen3.6-plus	Alibaba	Proprietary	1502.4±13.2	2.5k	$0.33	$1.9	1M
65	56–73	deepseek-v4-flash-high-preview	DeepSeek	MIT	1501.1±21.9	777	$0.30	$1.2	1M
66	58–71	deepseek-v4-pro	DeepSeek	MIT	1500.2±13.9	2.6k	$1.3	$4.0	1M
67	58–73	kimi-k2.5-thinking	Moonshot	Modified MIT	1496.3±17.2	1.2k	$0.60	$3	-
68	61–73	gpt-5.5 (codex-harness)	OpenAI	Proprietary	1485.2±15.1	1.6k	$5	$30	-
69	62–73	minimax-m2.7	MiniMax	Modified MIT	1478.3±17	1.2k	$0.30	$1.2	205K
70	62–74	Inkling Small	Thinky	Apache 2.0	1476.5±19	1.1k	$0.45	$1.2	524K
71	64–73	inkling	Thinky	Apache 2.0	1474.8±11.8	2.7k	$1	$4.0	524K
72	65–75	qwen3.5-397b-a17b	Alibaba	Apache 2.0	1467.6±15.3	1.6k	$0.50	$3.6	262K
73	65–78	gpt-5.4-mini-high	OpenAI	Proprietary	1461.1±24.9	572	$0.75	$4.5	400K
74	71–79	gemini-3.1-pro-preview	Google	Proprietary	1446.7±13.1	3.0k	$1	$6	1M
75	72–81	gpt-5.4	OpenAI	Proprietary	1434.1±23.3	692	$2.5	$15	1M
76	73–81	laguna-m.1	Poolside	Apache 2.0	1425.3±26.3	557	$0.20	$0.40	262K
77	73–81	solar-pro4	Upstage	Proprietary	1424.7±22.2	825	$0.09	$0.36	524K
78	73–81	gemini-3-flash (thinking-minimal)	Google	Proprietary	1420.1±23.9	724	$0.50	$3	1M
79	74–81	grok-4.20-beta-0309-reasoning	SpaceXAI	Proprietary	1416.4±17.8	1.7k	$1.3	$2.5	1M
80	75–81	claude-haiku-4-5-20251001	Anthropic	Proprietary	1405.6±14.6	2.2k	$1	$5	200K
81	75–82	gemma-4-31b	Google	Apache 2.0	1398.6±18.9	1.1k	$0.99	$1.5	131K
82	81–82	grok-4.3	SpaceXAI	Proprietary	1371.5±17.1	1.5k	$1.3	$2.5	1M

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 ↗