大模型排行榜
Arena人在盲测中对文本回答作出偏好选择。
Arena AI 原榜 · 榜单日期
| 排名 | 模型 | 评分 | 对战票数 |
|---|---|---|---|
| 1 | Gemini 4 Carbon (High)google | 1,5391530–1549 | 4,262 |
| 2 | Gemini 4 Argon (High)google | 1,5281523–1533 | 27,193 |
| 3 | Gemini 4 Neon (High)google | 1,5241515–1532 | 5,555 |
| 4 | Claude Opus 5.5 (High)anthropic | 1,5131507–1519 | 18,286 |
| 5 | Claude Fable 5.1 (Max)anthropic | 1,5121507–1518 | 22,408 |
| 6 | Claude Opus 5 (Max)anthropic | 1,5061502–1511 | 35,534 |
| 7 | Muse Wonder (Max)meta | 1,5031498–1509 | 16,158 |
| 8 | Claude Opus 5 (High)anthropic | 1,5021499–1506 | 73,414 |
| 9 | Gemini 4 Hydrogen (High)google | 1,5021493–1511 | 4,743 |
| 10 | Claude Opus 4.6 (High)anthropic | 1,5011498–1505 | 84,188 |
| 11 | Gemini 3.8 Flash (High)google | 1,5001496–1504 | 73,351 |
| 12 | Claude Opus 4.6anthropic | 1,4971494–1500 | 88,918 |
| 13 | Muse Spark 1.3 (Max)meta | 1,4941489–1498 | 30,142 |
| 14 | MiMo-V2.6-Proxiaomi | 1,4931483–1502 | 4,045 |
| 15 | Claude Fable 5 (High)anthropic | 1,4921488–1496 | 45,315 |
| 16 | Claude Opus 4.7 (High)anthropic | 1,4901486–1494 | 64,074 |
| 17 | Gemini 3.7 Flash (High)google | 1,4891485–1494 | 33,087 |
| 18 | Qwen3.8-Maxalibaba | 1,4851481–1490 | 40,248 |
| 19 | Muse Spark 1.2 (xHigh)meta | 1,4841479–1490 | 18,264 |
| 20 | Claude Opus 4.7anthropic | 1,4831480–1487 | 65,152 |
| 21 | Gemini 3.5 Flash (High)google | 1,4831479–1487 | 47,470 |
| 22 | Gemini 3.1 Pro Previewgoogle | 1,4801477–1483 | 131,889 |
| 23 | Gemini 3 Progoogle | 1,4801476–1483 | 41,792 |
| 24 | Gemini 3.6 Flash (High)google | 1,4801475–1484 | 35,275 |
| 25 | Muse Spark 1.1meta | 1,4791475–1483 | 45,342 |
| 26 | Claude Fable 5.2 (Max)anthropic | 1,4781470–1487 | 5,216 |
| 27 | Gemini 3.5 Flash (Medium)google | 1,4781474–1482 | 46,163 |
| 28 | GLM-5.3 (Max)zai | 1,4751471–1480 | 34,708 |
| 29 | Kimi K3 (Max)moonshot | 1,4751471–1479 | 44,364 |
| 30 | Qwen3.7-Max-Previewalibaba | 1,4751465–1484 | 3,888 |
| 31 | Muse Sparkmeta | 1,4741469–1480 | 13,972 |
| 32 | GLM-5.3-Flashzai | 1,4711467–1476 | 56,089 |
| 33 | Qwen3.5-Max-Previewalibaba | 1,4711466–1476 | 22,361 |
| 34 | GPT-5.5 (High)openai | 1,4711467–1475 | 68,109 |
| 35 | Claude Sonnet 5.5 (xHigh)anthropic | 1,4701464–1475 | 16,925 |
| 36 | GLM-5.2 (Max)zai | 1,4691465–1474 | 53,978 |
| 37 | GPT-5.4 (High)openai | 1,4691465–1473 | 63,928 |
| 38 | ERNIE 5.1baidu | 1,4681463–1472 | 38,932 |
| 39 | GPT-5.5openai | 1,4661463–1470 | 69,700 |
| 40 | Gemini 3 Flashgoogle | 1,4661462–1470 | 31,095 |
| 41 | MiMo-V2.5-Proxiaomi | 1,4641460–1467 | 76,511 |
| 42 | DeepSeek-V4.1-Flash-20260910 (Max)deepseek | 1,4631457–1469 | 13,492 |
| 43 | GLM-5.1zai | 1,4631459–1466 | 66,799 |
| 44 | Claude Opus 4.8 (High)anthropic | 1,4621458–1465 | 62,925 |
| 45 | Gemini 2.5 Progoogle | 1,4581455–1460 | 124,685 |
| 46 | Kimi K3.1 Pro (Max)moonshot | 1,4581450–1466 | 6,658 |
| 47 | Claude Sonnet 4.6anthropic | 1,4581454–1461 | 69,794 |
| 48 | GPT-5.6 Sol (xHigh)openai | 1,4571453–1462 | 43,782 |
| 49 | Kimi K2.6moonshot | 1,4551451–1460 | 39,440 |
| 50 | Qwen3.7-Plusalibaba | 1,4541450–1458 | 41,181 |