Arena 文本模型排行榜

查看 Arena 文本任务的公开名次、评分与投票数据。

评测来源 Arena · 文本原榜发布 查看原榜评分方法

Arena Score

421 个模型
Arena · 文本 · Arena Score
排名模型 / 开发机构Arena Score投票95% CI许可详情
01Gemini 4 Carbon (High)Google1526Arena Score投票4273.095% CI1517–1536许可Proprietary投票4273.095% CI1517–1536许可Proprietary详情
02Gemini 4 Argon (High)Google1522Arena Score投票20779.095% CI1516–1527许可Proprietary投票20779.095% CI1516–1527许可Proprietary详情
03Muse Wonder (Max)Meta1511Arena Score投票12696.095% CI1505–1517许可Proprietary投票12696.095% CI1505–1517许可Proprietary详情
04Claude Opus 5.5 (High)Anthropic1507Arena Score投票15793.095% CI1501–1513许可Proprietary投票15793.095% CI1501–1513许可Proprietary详情
05Claude Fable 5 (High)Anthropic1504Arena Score投票45469.095% CI1500–1508许可Proprietary投票45469.095% CI1500–1508许可Proprietary详情
06Claude Opus 4.6 (High)Anthropic1503Arena Score投票84375.095% CI1500–1506许可Proprietary投票84375.095% CI1500–1506许可Proprietary详情
07Claude Fable 5.1 (Max)Anthropic1502Arena Score投票20511.095% CI1497–1508许可Proprietary投票20511.095% CI1497–1508许可Proprietary详情
08Claude Opus 4.7 (High)Anthropic1501Arena Score投票64260.095% CI1497–1505许可Proprietary投票64260.095% CI1497–1505许可Proprietary详情
09Gemini 3.8 Flash (High)Google1497Arena Score投票62675.095% CI1492–1501许可Proprietary投票62675.095% CI1492–1501许可Proprietary详情
10Muse Spark 1.3 (Max)Meta1497Arena Score投票26806.095% CI1491–1502许可Proprietary投票26806.095% CI1491–1502许可Proprietary详情
11Claude Opus 4.6Anthropic1496Arena Score投票89177.095% CI1493–1500许可Proprietary投票89177.095% CI1493–1500许可Proprietary详情
12Claude Opus 4.7Anthropic1494Arena Score投票65309.095% CI1490–1498许可Proprietary投票65309.095% CI1490–1498许可Proprietary详情
13Muse Spark 1.2 (xHigh)Meta1493Arena Score投票15563.095% CI1487–1499许可Proprietary投票15563.095% CI1487–1499许可Proprietary详情
14Muse Spark 1.1Meta1491Arena Score投票45519.095% CI1487–1495许可Proprietary投票45519.095% CI1487–1495许可Proprietary详情
15Gemini 4 Hydrogen (High)Google1490Arena Score投票4741.095% CI1481–1499许可Proprietary投票4741.095% CI1481–1499许可Proprietary来源
16Claude Opus 5 (High)Anthropic1490Arena Score投票73632.095% CI1486–1493许可Proprietary投票73632.095% CI1486–1493许可Proprietary详情
17Gemini 3.7 Flash (High)Google1489Arena Score投票33251.095% CI1485–1494许可Proprietary投票33251.095% CI1485–1494许可Proprietary详情
18Muse SparkMeta1489Arena Score投票13985.095% CI1483–1495许可Proprietary投票13985.095% CI1483–1495许可Proprietary详情
19Claude Opus 5 (Max)Anthropic1489Arena Score投票35670.095% CI1484–1494许可Proprietary投票35670.095% CI1484–1494许可Proprietary详情
20Kimi K3 (Max)Moonshot AI1487Arena Score投票41402.095% CI1483–1492许可Kimi K3 license投票41402.095% CI1483–1492许可Kimi K3 license详情
21Gemini 3.1 Pro PreviewGoogle1487Arena Score投票132181.095% CI1484–1490许可Proprietary投票132181.095% CI1484–1490许可Proprietary详情
22Gemini 3 ProGoogle1486Arena Score投票41803.095% CI1482–1489许可Proprietary投票41803.095% CI1482–1489许可Proprietary详情
23GPT-5.6 Sol (xHigh)OpenAI1485Arena Score投票43925.095% CI1481–1489许可Proprietary投票43925.095% CI1481–1489许可Proprietary详情
24Kimi K3.1 Pro (Max)Moonshot AI1485Arena Score投票3995.095% CI1475–1495许可Kimi K3.1 Pro License投票3995.095% CI1475–1495许可Kimi K3.1 Pro License详情
25Qwen3.8-MaxAlibaba1484Arena Score投票36936.095% CI1480–1489许可Proprietary投票36936.095% CI1480–1489许可Proprietary详情
26Gemini 3.6 Flash (High)Google1483Arena Score投票35362.095% CI1478–1487许可Proprietary投票35362.095% CI1478–1487许可Proprietary详情
27Claude Opus 4.8 (High)Anthropic1482Arena Score投票63082.095% CI1478–1486许可Proprietary投票63082.095% CI1478–1486许可Proprietary详情
28GLM-5.3 (Max)Z.ai1482Arena Score投票31604.095% CI1477–1486许可MIT投票31604.095% CI1477–1486许可MIT详情
29GPT-5.5 (High)OpenAI1481Arena Score投票68294.095% CI1478–1485许可Proprietary投票68294.095% CI1478–1485许可Proprietary详情
30GPT-6.1 Sol (Max)OpenAI1481Arena Score投票13498.095% CI1475–1487许可Proprietary投票13498.095% CI1475–1487许可Proprietary详情
31MiMo-V2.6-ProXiaomi1480Arena Score投票4064.095% CI1470–1489许可MIT投票4064.095% CI1470–1489许可MIT详情
32Gemini 3.5 Flash (High)Google1478Arena Score投票47588.095% CI1474–1482许可Proprietary投票47588.095% CI1474–1482许可Proprietary详情
33GPT-5.5OpenAI1476Arena Score投票69859.095% CI1472–1480许可Proprietary投票69859.095% CI1472–1480许可Proprietary详情
34Gemini 3.5 Flash (Medium)Google1476Arena Score投票46329.095% CI1472–1480许可Proprietary投票46329.095% CI1472–1480许可Proprietary详情
35GPT-5.2 Chat (20260210)OpenAI1476Arena Score投票35669.095% CI1472–1480许可Proprietary投票35669.095% CI1472–1480许可Proprietary详情
36GLM-5.3-FlashZ.ai1475Arena Score投票51019.095% CI1471–1480许可MIT投票51019.095% CI1471–1480许可MIT详情
37GPT-5.4 (High)OpenAI1475Arena Score投票64067.095% CI1471–1479许可Proprietary投票64067.095% CI1471–1479许可Proprietary详情
38DeepSeek-V4.1-Flash-20260910 (Max)DeepSeek1475Arena Score投票13547.095% CI1469–1481许可MIT投票13547.095% CI1469–1481许可MIT详情
39GLM-5.2 (Max)Z.ai1475Arena Score投票54143.095% CI1471–1479许可MIT投票54143.095% CI1471–1479许可MIT详情
40Grok 4.20 Beta 1xAI1475Arena Score投票27653.095% CI1470–1479许可Proprietary投票27653.095% CI1470–1479许可Proprietary详情
41Claude Opus 4.8Anthropic1474Arena Score投票64116.095% CI1471–1478许可Proprietary投票64116.095% CI1471–1478许可Proprietary详情
42GPT-6 Astra (Max)OpenAI1474Arena Score投票18745.095% CI1468–1479许可Proprietary投票18745.095% CI1468–1479许可Proprietary详情
43Qwen3.7-Max-PreviewAlibaba1474Arena Score投票3896.095% CI1464–1484许可Proprietary投票3896.095% CI1464–1484许可Proprietary详情
44Claude Opus 4.5 (High, 32K, 20251101)Anthropic1474Arena Score投票37245.095% CI1470–1477许可Proprietary投票37245.095% CI1470–1477许可Proprietary详情
45GPT-5.5 InstantOpenAI1473Arena Score投票26859.095% CI1468–1478许可Proprietary投票26859.095% CI1468–1478许可Proprietary详情
46Gemini 3 FlashGoogle1473Arena Score投票31126.095% CI1469–1477许可Proprietary投票31126.095% CI1469–1477许可Proprietary详情
47Claude Sonnet 5.5 (xHigh)Anthropic1473Arena Score投票14355.095% CI1467–1479许可Proprietary投票14355.095% CI1467–1479许可Proprietary详情
48Claude Sonnet 4.6Anthropic1472Arena Score投票69975.095% CI1469–1476许可Proprietary投票69975.095% CI1469–1476许可Proprietary详情
49Grok 4.20 Beta (Reasoning, 0309)xAI1472Arena Score投票65690.095% CI1468–1476许可Proprietary投票65690.095% CI1468–1476许可Proprietary详情
50Grok 4.20 Multi-Agent Beta (0309)xAI1471Arena Score投票64196.095% CI1467–1475许可Proprietary投票64196.095% CI1467–1475许可Proprietary详情
显示更多已显示 50 / 421

数据来自 Arena 官方公开数据集(CC BY 4.0);保留原榜名次与评分。 官方数据集

未公布的数据记为 —