LMArena Math Arena 数学推理能力排行榜
基于 LMArena Math Arena 用户匿名投票的最新AI大模型数学推理能力排行榜，涵盖各模型的 Elo 得分、95% 置信区间、投票量、机构与许可证。
榜首模型
Gemini 4 Argon (High)
最高得分
1529.00
模型数量
403
数据版本
2026年10月20日
数据来源: LM Arena
关于本排行榜
本排行榜展示了当前 AI 大模型在数学推理任务中的实力排名。数据来源于 LMArena 的 Math 子赛道，通过真实用户匿名盲测投票评估各模型在数学解题任务中的表现。
评测方法概要
匿名盲测：用户提出数学题目后，由两个"隐藏身份"的模型分别作答，用户投票选出解题更优的一方，排除品牌偏见。
Elo 评分：采用 Bradley-Terry 模型计算 Elo 分数，分数越高说明该模型在数学场景中被用户更频繁地选择。
覆盖多种数学场景：包括代数、几何、计算推理、竞赛数学等多元化的真实数学任务。
DataLearner 在原始数据基础上提供中文解读与深度分析，并将排行榜模型关联至 DataLearner 模型库，方便您一键查看模型详情、API 定价、评测得分等完整信息。
排名总表
| 排名 | 模型名称 | 得分 | 95% CI | 投票数 | 机构 | 许可证 |
|---|---|---|---|---|---|---|
 Gemini 4 Argon (High)Google | 1529.00 | +/-34 | 273 | Proprietary | ||
 Claude Fable 5.1 (Max)Anthropic | 1527.00 | +/-21 | 746 | Anthropic | Proprietary | |
| Gemini 3.8 Flash (High)Google | 1526.00 | +/-13 | 2,094 | Proprietary | ||
| 4 | Claude Fable 5 (High)Anthropic | 1521.00 | +/-13 | 2,161 | Anthropic | Proprietary |
| 5 | Claude Opus 5 (high)Anthropic | 1521.00 | +/-11 | 3,315 | Anthropic | Proprietary |
| 6 | Claude Opus 5 (max)Anthropic | 1518.00 | +/-15 | 1,597 | Anthropic | Proprietary |
| 7 | Claude Opus 4.6 (high)Anthropic | 1517.00 | +/-10 | 4,248 | Anthropic | Proprietary |
| 8 | Claude Opus 5.5 (High)Anthropic | 1512.00 | +/-26 | 491 | Anthropic | Proprietary |
| 9 |  GPT-6.1 Sol (Max)OpenAI | 1512.00 | +/-28 | 390 | OpenAI | Proprietary |
| 10 | Gemini 3.7 Flash (high)Google DeepMind | 1507.00 | +/-15 | 1,501 | Google DeepMind | Proprietary |
| 11 | Gemini 3.6 Flash (High)Google | 1504.00 | +/-14 | 1,760 | Proprietary | |
| 12 |  Muse Spark 1.3 (Max)Meta | 1504.00 | +/-19 | 948 | Meta | Proprietary |
| 13 |  GLM-5.3-FlashZ.ai | 1504.00 | +/-14 | 1,820 | Z.ai | MIT |
| 14 | Claude Opus 4.6Anthropic | 1504.00 | +/-9 | 4,741 | Anthropic | Proprietary |
| 15 | Claude Opus 4.7 (High)Anthropic | 1504.00 | +/-11 | 3,293 | Anthropic | Proprietary |
| 16 | GPT-5.5OpenAI | 1500.00 | +/-10 | 3,518 | OpenAI | Proprietary |
| 17 | Claude Sonnet 5.5 (xHigh)Anthropic | 1500.00 | +/-28 | 416 | Anthropic | Proprietary |
| 18 |  Kimi K3 (max)Moonshot AI | 1500.00 | +/-15 | 1,535 | Moonshot AI | Kimi K3 license |
| 19 |  Qwen3.8-MaxAlibaba | 1499.00 | +/-15 | 1,481 | Alibaba | Proprietary |
| 20 | Muse Wonder (Max)Meta | 1498.00 | +/-33 | 277 | Meta | Proprietary |
| 21 | GLM-5.3 (max)智谱AI | 1498.00 | +/-18 | 1,023 | 智谱AI | MIT |
| 22 | Gemini 3.5 Flash (High)Google | 1497.00 | +/-12 | 2,388 | Proprietary | |
| 23 | GPT-5.6 Sol (xhigh)OpenAI | 1495.00 | +/-13 | 1,981 | OpenAI | Proprietary |
| 24 |  DeepSeek-V4.1-Flash-20260910 (Max)DeepSeek | 1494.00 | +/-22 | 704 | DeepSeek | MIT |
| 25 | Muse Spark 1.1Facebook AI研究实验室 | 1494.00 | +/-13 | 2,076 | Facebook AI研究实验室 | Proprietary |
| 26 | Claude Opus 4.8 (High)Anthropic | 1494.00 | +/-11 | 2,974 | Anthropic | Proprietary |
| 27 | GPT-5.5 (high)OpenAI | 1493.00 | +/-10 | 3,475 | OpenAI | Proprietary |
| 28 | Muse Spark 1.2 (xhigh)Facebook AI研究实验室 | 1491.00 | +/-26 | 484 | Facebook AI研究实验室 | Proprietary |
| 29 | Opus 4.7Anthropic | 1491.00 | +/-11 | 3,389 | Anthropic | Proprietary |
| 30 | GPT-5.4 (High)OpenAI | 1491.00 | +/-11 | 3,377 | OpenAI | Proprietary |
| 31 | Gemini 3.1 Pro PreviewGoogle DeepMind | 1489.00 | +/-8 | 6,734 | Google DeepMind | Proprietary |
| 32 | MiMo-V2.6-ProXiaomi | 1489.00 | +/-40 | 215 | Xiaomi | MIT |
| 33 | GPT-6 Astra (max)OpenAI | 1487.00 | +/-22 | 687 | OpenAI | Proprietary |
| 34 | GLM-5.2 (max)智谱AI | 1487.00 | +/-12 | 2,303 | 智谱AI | MIT |
| 35 | Qwen3.7-Max-Preview阿里巴巴 | 1484.00 | +/-38 | 233 | 阿里巴巴 | Proprietary |
| 36 | Gemini 3.5 Flash (medium)Google DeepMind | 1482.00 | +/-13 | 2,214 | Google DeepMind | Proprietary |
| 37 | GPT-5.6 Terra (xhigh)OpenAI | 1481.00 | +/-13 | 2,056 | OpenAI | Proprietary |
| 38 |  Hy3腾讯AI实验室 | 1481.00 | +/-22 | 689 | 腾讯AI实验室 | Apache 2.0 |
| 39 | MiMo-V2.5-Pro小米 | 1479.00 | +/-10 | 3,709 | 小米 | MIT |
| 40 | Kimi K2.6Moonshot AI | 1479.00 | +/-13 | 2,072 | Moonshot AI | Modified MIT |
| 41 | GPT-5.6 Luna (xhigh)OpenAI | 1478.00 | +/-14 | 1,926 | OpenAI | Proprietary |
| 42 |  ERNIE 5.1Baidu | 1478.00 | +/-13 | 1,981 | Baidu | Proprietary |
| 43 | InklingThinking Machines Lab | 1478.00 | +/-13 | 1,951 | Thinking Machines Lab | Apache 2.0 |
| 44 | Gemini 3 ProGoogle | 1477.00 | +/-11 | 2,770 | Proprietary | |
| 45 | Claude Opus 4.8Anthropic | 1477.00 | +/-11 | 3,089 | Anthropic | Proprietary |
| 46 | Qwen3.6-Max-Preview阿里巴巴 | 1475.00 | +/-29 | 378 | 阿里巴巴 | Proprietary |
| 47 | GLM 5.1智谱AI | 1475.00 | +/-11 | 3,182 | 智谱AI | MIT |
| 48 | Gemini 3.0 FlashGoogle DeepMind | 1475.00 | +/-13 | 2,071 | Google DeepMind | Proprietary |
| 49 |  Grok 4.5xAI | 1473.00 | +/-14 | 1,765 | xAI | Proprietary |
| 50 | Claude Sonnet 5 (High)Anthropic | 1473.00 | +/-12 | 2,325 | Anthropic | Proprietary |
| 51 | Beam 1.0 501B A23BReflection | 1472.00 | +/-37 | 233 | Reflection | Apache 2.0 |
| 52 | Claude Opus 4.5 (High, 32K, 20251101)Anthropic | 1472.00 | +/-12 | 2,346 | Anthropic | Proprietary |
| 53 | Kimi K2.5 (thinking)Moonshot AI | 1471.00 | +/-9 | 4,174 | Moonshot AI | Modified MIT |
| 54 | MiMo-V2.6-FlashXiaomi | 1471.00 | +/-22 | 676 | Xiaomi | MIT |
| 55 | Gemma 4 26B A4BGoogle DeepMind | 1469.00 | +/-28 | 390 | Google DeepMind | Apache 2.0 |
| 56 |  Mistral Large 4Mistral | 1468.00 | +/-26 | 488 | Mistral | Proprietary |
| 57 | Qwen3.5 Max Preview阿里巴巴 | 1468.00 | +/-16 | 1,425 | 阿里巴巴 | Proprietary |
| 58 | Gemma 4 31BGoogle DeepMind | 1468.00 | +/-26 | 428 | Google DeepMind | Apache 2.0 |
| 59 | Muse SparkFacebook AI研究实验室 | 1466.00 | +/-20 | 891 | Facebook AI研究实验室 | Proprietary |
| 60 | Qwen3.7-Plus阿里巴巴 | 1466.00 | +/-14 | 1,917 | 阿里巴巴 | Proprietary |
| 61 | DeepSeek-V4-Pro Preview (High)DeepSeek | 1466.00 | +/-12 | 2,716 | DeepSeek | MIT |
| 62 |  Step 5 PreviewStepFun | 1466.00 | +/-27 | 443 | StepFun | Proprietary |
| 63 | Grok 4.20 Beta (Reasoning, 0309)SpaceXAI | 1465.00 | +/-10 | 3,536 | SpaceXAI | Proprietary |
| 64 | Claude Sonnet 4.6Anthropic | 1463.00 | +/-10 | 3,796 | Anthropic | Proprietary |
| 65 | Claude Opus 4.5 (20251101)Anthropic | 1463.00 | +/-9 | 4,544 | Anthropic | Proprietary |
| 66 | GPT-5.5 InstantOpenAI | 1460.00 | +/-16 | 1,530 | OpenAI | Proprietary |
| 67 | GPT-5.4OpenAI | 1460.00 | +/-10 | 3,557 | OpenAI | Proprietary |
| 68 | GPT-5.1 (high)OpenAI | 1458.00 | +/-12 | 2,580 | OpenAI | Proprietary |
| 69 | DeepSeek-V4-Pro (High, 20260813)DeepSeek | 1457.00 | +/-28 | 407 | DeepSeek | MIT |
| 70 | GPT-5.2 (high)OpenAI | 1456.00 | +/-11 | 3,125 | OpenAI | Proprietary |
| 71 | Claude Sonnet 4.5 (High, 32K, 20250929)Anthropic | 1455.00 | +/-8 | 5,075 | Anthropic | Proprietary |
| 72 | Grok 4.20 Multi-Agent Beta (0309)SpaceXAI | 1454.00 | +/-10 | 3,388 | SpaceXAI | Proprietary |
| 73 | Gemini 3 Flash (Minimal)Google | 1454.00 | +/-9 | 5,000 | Proprietary | |
| 74 | GPT-5.2 Chat (20260210)OpenAI | 1454.00 | +/-13 | 2,166 | OpenAI | Proprietary |
| 75 | Qwen 3.6 Plus Preview阿里巴巴 | 1453.00 | +/-12 | 2,685 | 阿里巴巴 | Proprietary |
| 76 | GPT-6 Sol (Max)OpenAI | 1452.00 | +/-21 | 767 | OpenAI | Proprietary |
| 77 | Inkling-SmallThinky | 1452.00 | +/-15 | 1,556 | Thinky | Apache 2.0 |
| 78 | Qwen3.5-397B-A17B阿里巴巴 | 1451.00 | +/-9 | 5,233 | 阿里巴巴 | Apache 2.0 |
| 79 | Qwen3.8-27BAlibaba | 1450.00 | +/-17 | 1,250 | Alibaba | Apache 2.0 |
| 80 | MiMo-V2-ProXiaomi | 1450.00 | +/-15 | 1,671 | Xiaomi | Proprietary |
| 81 |  Dola Seed 2.0 ProBytedance | 1449.00 | +/-10 | 4,348 | Bytedance | Proprietary |
| 82 | Grok 4.20 Beta 1SpaceXAI | 1448.00 | +/-14 | 1,684 | SpaceXAI | Proprietary |
| 83 | o3 (2025-04-16)OpenAI | 1448.00 | +/-10 | 3,701 | OpenAI | Proprietary |
| 84 | GPT-6 Luna (Max)OpenAI | 1447.00 | +/-11 | 3,477 | OpenAI | Proprietary |
| 85 | Muse GlimmerMeta | 1445.00 | +/-38 | 226 | Meta | Apache-2.0 |
| 86 | Grok 4.6 (high)xAI | 1445.00 | +/-15 | 1,521 | xAI | Proprietary |
| 87 | DeepSeek-V4-ProDeepSeek-AI | 1444.00 | +/-11 | 3,068 | DeepSeek-AI | MIT |
| 88 |  NVIDIA-Nemotron-3-Ultra-550B-A55B (NVFP4)Nvidia | 1444.00 | +/-24 | 580 | Nvidia | OpenMDW-1.1 |
| 89 | Claude Opus 4.1 (Thinking 16K, 20250805)Anthropic | 1444.00 | +/-10 | 3,058 | Anthropic | Proprietary |
| 90 | Grok 4.7 (xHigh)SpaceXAI | 1444.00 | +/-22 | 693 | SpaceXAI | Proprietary |
| 91 | GLM-5智谱AI | 1443.00 | +/-14 | 1,694 | 智谱AI | MIT |
| 92 | Gemini 3.5 Flash-LiteGoogle DeepMind | 1443.00 | +/-15 | 1,622 | Google DeepMind | Proprietary |
| 93 | Grok 4.1 (Thinking)SpaceXAI | 1442.00 | +/-10 | 3,948 | SpaceXAI | Proprietary |
| 94 | DeepSeek-V4-Flash Preview (High)DeepSeek | 1442.00 | +/-12 | 2,624 | DeepSeek | MIT |
| 95 | Kimi K2.5 (Instant)Moonshot | 1441.00 | +/-24 | 556 | Moonshot | Modified MIT |
| 96 | GPT-5.4 mini (High)OpenAI | 1439.00 | +/-11 | 3,369 | OpenAI | Proprietary |
| 97 | Gemini 2.5 ProGoogle | 1439.00 | +/-7 | 7,818 | Proprietary | |
| 98 | GLM-5V-Turbo智谱AI | 1438.00 | +/-25 | 485 | 智谱AI | Proprietary |
| 99 | MiMo-V2.5小米 | 1438.00 | +/-12 | 2,461 | 小米 | MIT |
| 100 | Kimi K2 Thinking TurboMoonshot | 1436.00 | +/-10 | 3,922 | Moonshot | Modified MIT |
| 101 | Gemini 3.1 Flash-LiteGoogle DeepMind | 1436.00 | +/-10 | 3,619 | Google DeepMind | Proprietary |
| 102 | GPT-5 (high)OpenAI | 1435.00 | +/-14 | 1,869 | OpenAI | Proprietary |
| 103 | LongCat-Flash-Chat Exp (2602)Meituan | 1434.00 | +/-14 | 1,827 | Meituan | Proprietary |
| 104 | ERNIE 5.0 (0110)Baidu | 1434.00 | +/-12 | 2,256 | Baidu | Proprietary |
| 105 | MiMo-V2-OmniXiaomi | 1434.00 | +/-19 | 980 | Xiaomi | Proprietary |
| 106 | Qwen3 Max (Preview)阿里巴巴 | 1433.00 | +/-15 | 1,495 | 阿里巴巴 | Proprietary |
| 107 | Claude Opus 4.1 (20250805)Anthropic | 1433.00 | +/-8 | 4,807 | Anthropic | Proprietary |
| 108 | GPT-5.2OpenAI | 1432.00 | +/-9 | 4,642 | OpenAI | Proprietary |
| 109 | GPT-5.3 ChatOpenAI | 1431.00 | +/-13 | 2,139 | OpenAI | Proprietary |
| 110 |  MiniMax M3MiniMaxAI | 1430.00 | +/-11 | 3,307 | MiniMaxAI | MiniMax Community License |
| 111 | Hy3 previewTencent | 1430.00 | +/-27 | 430 | Tencent | tencent-hunyuan-community |
| 112 | DeepSeek-V3.2 (Thinking)DeepSeek | 1429.00 | +/-11 | 2,633 | DeepSeek | MIT |
| 113 | GLM-4.7智谱AI | 1428.00 | +/-20 | 738 | 智谱AI | MIT |
| 114 | Claude Sonnet 4.5 (20250929)Anthropic | 1427.00 | +/-8 | 5,124 | Anthropic | Proprietary |
| 115 | Mistral Medium 3.5MistralAI | 1427.00 | +/-24 | 565 | MistralAI | Modified MIT |
| 116 | DeepSeek V3.2DeepSeek-AI | 1426.00 | +/-11 | 3,140 | DeepSeek-AI | MIT |
| 117 | Grok 4.1SpaceXAI | 1426.00 | +/-9 | 4,359 | SpaceXAI | Proprietary |
| 118 | Grok 4 (0709)SpaceXAI | 1425.00 | +/-12 | 2,251 | SpaceXAI | Proprietary |
| 119 | Qwen3.5-122B-A10B阿里巴巴 | 1425.00 | +/-14 | 1,842 | 阿里巴巴 | Apache 2.0 |
| 120 | Grok 4 Fast (Non-Reasoning)SpaceXAI | 1425.00 | +/-30 | 381 | SpaceXAI | Proprietary |
| 121 | DeepSeek-V4-FlashDeepSeek-AI | 1424.00 | +/-12 | 2,699 | DeepSeek-AI | MIT |
| 122 | GPT-5.4 nano (High)OpenAI | 1424.00 | +/-11 | 3,261 | OpenAI | Proprietary |
| 123 | Qwen3-Max (2025-09-23)Alibaba | 1424.00 | +/-24 | 571 | Alibaba | Proprietary |
| 124 | Qwen3.5-27B阿里巴巴 | 1423.00 | +/-14 | 1,745 | 阿里巴巴 | Apache 2.0 |
| 125 |  amazon-nova-experimental-chat (26-02-10)Amazon | 1423.00 | +/-39 | 209 | Amazon | Proprietary |
| 126 | GPT-5.1OpenAI | 1423.00 | +/-11 | 2,965 | OpenAI | Proprietary |
| 127 | DeepSeek-V3.2-Exp (Thinking)DeepSeek | 1423.00 | +/-26 | 477 | DeepSeek | MIT |
| 128 | MiniMax-M2.7MiniMaxAI | 1422.00 | +/-9 | 4,372 | MiniMaxAI | Modified MIT |
| 129 | Claude Opus 4 (Thinking 16K, 20250514)Anthropic | 1421.00 | +/-12 | 2,230 | Anthropic | Proprietary |
| 130 | Grok 4.1 Fast (reasoning)xAI | 1421.00 | +/-10 | 3,607 | xAI | Proprietary |
| 131 | Solar Pro 4Upstage | 1419.00 | +/-17 | 1,181 | Upstage | Proprietary |
| 132 | GLM-4.6智谱AI | 1418.00 | +/-13 | 2,154 | 智谱AI | MIT |
| 133 | Grok 4.3 BetaxAI | 1418.00 | +/-11 | 3,418 | xAI | Proprietary |
| 134 | o4-mini (2025-04-16)OpenAI | 1418.00 | +/-11 | 2,916 | OpenAI | Proprietary |
| 135 | Qwen3-235B-A22B-Instruct (2507)Alibaba | 1417.00 | +/-8 | 6,081 | Alibaba | Apache 2.0 |
| 136 | Kimi K2 Preview (0905)Moonshot | 1416.00 | +/-21 | 755 | Moonshot | Modified MIT |
| 137 | LongCat-Flash-ChatMeituan | 1415.00 | +/-22 | 686 | Meituan | MIT |
| 138 | Qwen3-Next阿里巴巴 | 1414.00 | +/-17 | 1,204 | 阿里巴巴 | Apache 2.0 |
| 139 | DeepSeek-V3.1 (thinking)DeepSeek-AI | 1414.00 | +/-22 | 658 | DeepSeek-AI | MIT |
| 140 | DeepSeek-V3.1DeepSeek-AI | 1413.00 | +/-18 | 980 | DeepSeek-AI | MIT |
| 141 | DeepSeek-R1DeepSeek-AI | 1412.00 | +/-14 | 1,606 | DeepSeek-AI | MIT |
| 142 | amazon-nova-experimental-chat (26-01-10)Amazon | 1411.00 | +/-33 | 267 | Amazon | Proprietary |
| 143 | GLM-4.5智谱AI | 1411.00 | +/-16 | 1,401 | 智谱AI | MIT |
| 144 | Grok 4 Fast (reasoning)xAI | 1411.00 | +/-18 | 1,078 | xAI | Proprietary |
| 145 | ERNIE-5.0-Preview (1203)Baidu | 1411.00 | +/-23 | 630 | Baidu | Proprietary |
| 146 | GPT-5 ChatOpenAI | 1410.00 | +/-14 | 1,780 | OpenAI | Proprietary |
| 147 | DeepSeek V3.2-ExpDeepSeek-AI | 1410.00 | +/-20 | 792 | DeepSeek-AI | MIT |
| 148 | o1 (2024-12-17)OpenAI | 1409.00 | +/-11 | 2,986 | OpenAI | Proprietary |
| 149 | Qwen3-VL-235B-A22B-Instruct阿里巴巴 | 1409.00 | +/-23 | 699 | 阿里巴巴 | Apache 2.0 |
| 150 | Gemini 2.5 Flash Preview (09-2025)Google | 1408.00 | +/-13 | 1,978 | Proprietary | |
| 151 | GPT-4.5 Preview (2025-02-27)OpenAI | 1408.00 | +/-15 | 1,393 | OpenAI | Proprietary |
| 152 | Step 3.5 FlashStepFunAI | 1407.00 | +/-10 | 3,424 | StepFunAI | Apache 2.0 |
| 153 | NVIDIA-Nemotron-3.5-Super-VL-120B-A12B (NVFP4)Nvidia | 1407.00 | +/-38 | 222 | Nvidia | OpenMDW-1.1 |
| 154 | GPT-5-mini (high)OpenAI | 1406.00 | +/-15 | 1,458 | OpenAI | Proprietary |
| 155 | Gemini 2.5 FlashGoogle DeepMind | 1405.00 | +/-7 | 8,004 | Google DeepMind | Proprietary |
| 156 | o3-mini (High)OpenAI | 1405.00 | +/-13 | 1,909 | OpenAI | Proprietary |
| 157 | Claude Sonnet 4 (Thinking 32K, 20250514)Anthropic | 1404.00 | +/-13 | 2,015 | Anthropic | Proprietary |
| 158 | ChatGPT-4o Latest (20250326)OpenAI | 1403.00 | +/-8 | 5,784 | OpenAI | Proprietary |
| 159 | Hunyuan T1 (20250711)Tencent | 1403.00 | +/-38 | 233 | Tencent | Proprietary |
| 160 | Claude Opus 4 (20250514)Anthropic | 1403.00 | +/-11 | 2,744 | Anthropic | Proprietary |
| 161 | Claude Haiku 4.5 (20251001)Anthropic | 1402.00 | +/-7 | 8,434 | Anthropic | Proprietary |
| 162 | Step 3.5 FlashStepFunAI | 1401.00 | +/-11 | 3,388 | StepFunAI | Proprietary |
| 163 | Mistral Large 3MistralAI | 1401.00 | +/-9 | 4,893 | MistralAI | Apache 2.0 |
| 164 | Qwen3-VL-235B-A22B-Instruct (thinking)阿里巴巴 | 1399.00 | +/-28 | 424 | 阿里巴巴 | Apache 2.0 |
| 165 | Qwen3-32B阿里巴巴 | 1398.00 | +/-30 | 316 | 阿里巴巴 | Apache 2.0 |
| 166 | Qwen3.5-35B-A3B阿里巴巴 | 1398.00 | +/-14 | 1,850 | 阿里巴巴 | Apache 2.0 |
| 167 | amazon-nova-experimental-chat (12-10)Amazon | 1397.00 | +/-36 | 239 | Amazon | Proprietary |
| 168 | DeepSeek-V3.1 TerminusDeepSeek-AI | 1397.00 | +/-39 | 216 | DeepSeek-AI | MIT |
| 169 | Qwen3-235B-A22B-Thinking (2507)Alibaba | 1396.00 | +/-25 | 479 | Alibaba | Apache 2.0 |
| 170 | Mistral Medium 3.1 (2508)Mistral | 1395.00 | +/-8 | 5,994 | Mistral | Proprietary |
| 171 | amazon-nova-experimental-chat (11-10)Amazon | 1394.00 | +/-14 | 1,666 | Amazon | Proprietary |
| 172 | ERNIE-5.0-Preview (1022)Baidu | 1393.00 | +/-34 | 273 | Baidu | Proprietary |
| 173 | MiniMax M2.5MiniMaxAI | 1393.00 | +/-12 | 2,551 | MiniMaxAI | Modified MIT |
| 174 | DeepSeek-R1 (0528)DeepSeek | 1393.00 | +/-20 | 861 | DeepSeek | MIT |
| 175 | Qwen3-Next (thinking)阿里巴巴 | 1392.00 | +/-20 | 822 | 阿里巴巴 | Apache 2.0 |
| 176 | Qwen3-235B-A22B阿里巴巴 | 1392.00 | +/-14 | 1,604 | 阿里巴巴 | Apache 2.0 |
| 177 | Qwen3-235B-A22B (Non-Thinking)Alibaba | 1391.00 | +/-12 | 2,379 | Alibaba | Apache 2.0 |
| 178 | amazon-nova-experimental-chat (10-20)Amazon | 1391.00 | +/-20 | 815 | Amazon | Proprietary |
| 179 | MiniMax-M2.1 PreviewMiniMax | 1390.00 | +/-18 | 1,044 | MiniMax | MIT |
| 180 | Claude Sonnet 4 (20250514)Anthropic | 1390.00 | +/-12 | 2,448 | Anthropic | Proprietary |
| 181 | Grok 3 Mini (High)SpaceXAI | 1389.00 | +/-18 | 968 | SpaceXAI | Proprietary |
| 182 | Kimi K2 Preview (0711)Moonshot | 1388.00 | +/-14 | 1,697 | Moonshot | Modified MIT |
| 183 | Llama 3.3 Nemotron Super 49B v1.5Nvidia | 1387.00 | +/-39 | 194 | Nvidia | Nvidia Open |
| 184 | OpenAI o1OpenAI | 1387.00 | +/-10 | 4,569 | OpenAI | Proprietary |
| 185 | GLM-4.5-Air智谱AI | 1386.00 | +/-15 | 1,528 | 智谱AI | MIT |
| 186 | Claude 3.7 Sonnet (Thinking 32K, 20250219)Anthropic | 1386.00 | +/-11 | 2,786 | Anthropic | Proprietary |
| 187 | Trinity Large Thinking Apache 2.0 | 1382.00 | +/-15 | 1,723 | — | — |
| 188 | OpenAI o3-miniOpenAI | 1382.00 | +/-9 | 4,700 | OpenAI | Proprietary |
| 189 | GPT OSS 120BOpenAI | 1381.00 | +/-14 | 1,781 | OpenAI | Apache 2.0 |
| 190 | Llama 3.1 Nemotron Ultra 253B v1Nvidia | 1381.00 | +/-37 | 209 | Nvidia | Nvidia Open Model |
| 191 | Qwen3-30B-A3B-Instruct (2507)Alibaba | 1378.00 | +/-15 | 1,394 | Alibaba | Apache 2.0 |
| 192 | MiMo-V2-Flash (Non-Thinking)Xiaomi | 1376.00 | +/-11 | 2,981 | Xiaomi | MIT |
| 193 | Qwen3-Coder-480B-A35B阿里巴巴 | 1376.00 | +/-15 | 1,610 | 阿里巴巴 | Apache 2.0 |
| 194 | GPT-4.1 (2025-04-14)OpenAI | 1375.00 | +/-10 | 3,227 | OpenAI | Proprietary |
| 195 | NVIDIA-Nemotron-3.5-Lightning-30B-A3B (NVFP4)Nvidia | 1374.00 | +/-20 | 954 | Nvidia | OpenMDW-1.1 |
| 196 | INTELLECT-3 MIT | 1373.00 | +/-30 | 343 | — | — |
| 197 | Grok 3 Preview (02-24)SpaceXAI | 1372.00 | +/-11 | 2,682 | SpaceXAI | Proprietary |
| 198 | MiMo-V2-Flash (Thinking)Xiaomi | 1372.00 | +/-22 | 660 | Xiaomi | MIT |
| 199 | NVIDIA-Nemotron-3-Super-120B-A12BNvidia | 1371.00 | +/-24 | 547 | Nvidia | NVIDIA Open Model |
| 200 | Grok 3 Mini BetaSpaceXAI | 1370.00 | +/-14 | 1,513 | SpaceXAI | Proprietary |
| 201 | MiniMax-M1MiniMax | 1370.00 | +/-14 | 1,771 | MiniMax | Apache 2.0 |
| 202 | DeepSeek-V3 (0324)DeepSeek | 1367.00 | +/-10 | 3,165 | DeepSeek | MIT |
| 203 | Gemini 2.5 Flash-Lite Preview (Thinking, 06-17)Google | 1363.00 | +/-13 | 2,073 | Proprietary | |
| 204 | Claude 3.7 Sonnet (20250219)Anthropic | 1363.00 | +/-10 | 3,350 | Anthropic | Proprietary |
| 205 | QwQ-32B阿里巴巴 | 1363.00 | +/-14 | 1,712 | 阿里巴巴 | Apache 2.0 |
| 206 | GLM-4.7-Flash智谱AI | 1362.00 | +/-21 | 756 | 智谱AI | MIT |
| 207 | Qwen2.5-Max阿里巴巴 | 1362.00 | +/-10 | 3,308 | 阿里巴巴 | Proprietary |
| 208 | Step3StepFunAI | 1362.00 | +/-31 | 349 | StepFunAI | Apache 2.0 |
| 209 | OpenAI o1-miniOpenAI | 1362.00 | +/-8 | 7,499 | OpenAI | Proprietary |
| 210 | Trinity Large Preview Apache 2.0 | 1362.00 | +/-14 | 1,989 | — | — |
| 211 | Gemini 2.5 Flash-Lite Preview (Non-Thinking, 09-2025)Google | 1362.00 | +/-11 | 2,954 | Proprietary | |
| 212 | GLM-4.5V智谱AI | 1360.00 | +/-34 | 269 | 智谱AI | MIT |
| 213 | Gemini 2.0 Flash (001)Google | 1356.00 | +/-9 | 4,060 | Proprietary | |
| 214 | GPT-4.1 mini (2025-04-14)OpenAI | 1353.00 | +/-11 | 2,674 | OpenAI | Proprietary |
| 215 | MiniMax M2MiniMaxAI | 1352.00 | +/-33 | 320 | MiniMaxAI | Apache 2.0 |
| 216 | Claude 3.5 Sonnet (20241022)Anthropic | 1352.00 | +/-7 | 10,012 | Anthropic | Proprietary |
| 217 | Qwen3-30B-A3B阿里巴巴 | 1351.00 | +/-14 | 1,706 | 阿里巴巴 | Apache 2.0 |
| 218 |  Phi-3-medium 14B-preview微软 | 1348.00 | +/-12 | 2,233 | 微软 | Proprietary |
| 219 | NVIDIA-Nemotron-3-Nano-30B-A3B (BF16)Nvidia | 1348.00 | +/-18 | 1,014 | Nvidia | NVIDIA Open Model |
| 220 | GPT-5 nano (High)OpenAI | 1346.00 | +/-27 | 494 | OpenAI | Proprietary |
| 221 | Hunyuan TurboS (20250416)Tencent | 1345.00 | +/-19 | 853 | Tencent | Proprietary |
| 222 | Ling-flash-2.0 AntGroup | 1345.00 | +/-27 | 452 | Group | MIT |
| 223 | Claude 3.5 Sonnet (20240620)Anthropic | 1342.00 | +/-8 | 11,359 | Anthropic | Proprietary |
| 224 | Gemini 1.5 Pro (002)Google | 1340.00 | +/-7 | 7,610 | Proprietary | |
| 225 | Ring-flash-2.0 AntGroup | 1339.00 | +/-27 | 447 | Group | MIT |
| 226 | Mistral Small 3.2 (2506)Mistral | 1337.00 | +/-17 | 1,049 | Mistral | Apache 2.0 |
| 227 | gpt-oss-20bOpenAI | 1335.00 | +/-22 | 678 | OpenAI | Apache 2.0 |
| 228 | Amazon Nova 2 LiteAmazon | 1330.00 | +/-20 | 861 | Amazon | Proprietary |
| 229 | Gemini 2.0 Flash-Lite Preview (02-05)Google | 1327.00 | +/-10 | 2,814 | Proprietary | |
| 230 | Gemma 3 27B ITGoogle | 1323.00 | +/-10 | 3,587 | Gemma | |
| 231 | Qwen-Plus (0125)Alibaba | 1322.00 | +/-19 | 732 | Alibaba | Proprietary |
| 232 | Granite 4.1 8BIBM | 1319.00 | +/-39 | 248 | IBM | Apache 2.0 |
| 233 | Llama 3.1 405B Instruct (FP8)Meta | 1319.00 | +/-8 | 8,482 | Meta | Llama 3.1 Community |
| 234 | Gemma 3 12B ITGoogle | 1318.00 | +/-27 | 389 | Gemma | |
| 235 | Llama 4 Maverick 17B 128E InstructMeta | 1317.00 | +/-11 | 2,825 | Meta | Llama 4 |
| 236 | Llama 3.1 405B Instruct (BF16)Meta | 1316.00 | +/-8 | 5,215 | Meta | Llama 3.1 Community |
| 237 | Claude 3 Opus (20240229)Anthropic | 1312.00 | +/-7 | 25,769 | Anthropic | Proprietary |
| 238 | Step-2 16K Exp (202412)StepFun | 1312.00 | +/-20 | 642 | StepFun | Proprietary |
| 239 | Olmo 3 32B ThinkAi2 | 1312.00 | +/-31 | 329 | Ai2 | Apache 2.0 |
| 240 | Athene V2 Chat NexusFlow | 1311.00 | +/-10 | 3,412 | — | — |
| 241 |  Command A (03-2025)Cohere | 1310.00 | +/-9 | 3,993 | Cohere | CC-BY-NC-4.0 |
| 242 | DeepSeek-V3DeepSeek-AI | 1310.00 | +/-11 | 2,721 | DeepSeek-AI | DeepSeek |
| 243 | Llama 4 Scout 17B 16E InstructMeta | 1309.00 | +/-13 | 1,942 | Meta | Llama |
| 244 | GPT-4o (2024-08-06)OpenAI | 1309.00 | +/-8 | 6,826 | OpenAI | Proprietary |
| 245 | Gemini Advanced (0514)Google | 1308.00 | +/-10 | 6,395 | Proprietary | |
| 246 | GPT-4o (2024-05-13)OpenAI | 1305.00 | +/-7 | 15,103 | OpenAI | Proprietary |
| 247 | Yi Lightning Proprietary | 1304.00 | +/-10 | 3,921 | — | — |
| 248 | Olmo 3.1 32B ThinkAi2 | 1303.00 | +/-26 | 478 | Ai2 | Apache 2.0 |
| 249 | Qwen2.5-Plus (1127)Alibaba | 1303.00 | +/-14 | 1,404 | Alibaba | Proprietary |
| 250 | GPT-4 Turbo Preview (1106)OpenAI | 1303.00 | +/-8 | 13,306 | OpenAI | Proprietary |
| 251 | Hunyuan TurboS (20250226)Tencent | 1301.00 | +/-31 | 238 | Tencent | Proprietary |
| 252 | Gemini 1.5 Pro (001)Google | 1300.00 | +/-8 | 10,492 | Proprietary | |
| 253 | GPT-4 Turbo Preview (0125)OpenAI | 1299.00 | +/-8 | 12,374 | OpenAI | Proprietary |
| 254 | GLM-4-Plus (0111)Z.ai | 1298.00 | +/-19 | 721 | Z.ai | Proprietary |
| 255 | GPT-4 Turbo (2024-04-09)OpenAI | 1296.00 | +/-8 | 13,217 | OpenAI | Proprietary |
| 256 | Olmo 3.1 32B InstructAi2 | 1296.00 | +/-23 | 693 | Ai2 | Apache 2.0 |
| 257 | Qwen2.5-VL-72B-Instruct阿里巴巴 | 1296.00 | +/-8 | 5,415 | 阿里巴巴 | Qwen |
| 258 | Step-1o Turbo (202506)StepFun | 1296.00 | +/-24 | 564 | StepFun | Proprietary |
| 259 | Llama 3.3 70B InstructMeta | 1296.00 | +/-8 | 5,779 | Meta | Llama-3.3 |
| 260 | Grok 2 (2024-08-13)SpaceXAI | 1294.00 | +/-7 | 8,950 | SpaceXAI | Proprietary |
| 261 | Hunyuan Large (2025-02-10)Tencent | 1293.00 | +/-24 | 497 | Tencent | Proprietary |
| 262 | DeepSeek-V2.5 (1210)DeepSeek | 1292.00 | +/-17 | 1,031 | DeepSeek | DeepSeek |
| 263 | Qwen-Max (0919)Alibaba | 1291.00 | +/-12 | 2,249 | Alibaba | Qwen |
| 264 | Hunyuan Standard (2025-02-10)Tencent | 1290.00 | +/-24 | 499 | Tencent | Proprietary |
| 265 | Gemini 1.5 Flash (002)Google | 1289.00 | +/-9 | 4,789 | Proprietary | |
| 266 | Mistral Large 2 (2407)Mistral | 1288.00 | +/-8 | 6,664 | Mistral | Mistral Research |
| 267 | DeepSeek V2.5DeepSeek-AI | 1287.00 | +/-10 | 3,649 | DeepSeek-AI | DeepSeek |
| 268 | Magistral Medium (2506)Mistral | 1287.00 | +/-26 | 551 | Mistral | Proprietary |
| 269 | GLM-4-PlusZhipu AI | 1287.00 | +/-10 | 3,599 | Zhipu AI | Proprietary |
| 270 | Claude 3.5 Haiku (20241022)Anthropic | 1287.00 | +/-8 | 6,333 | Anthropic | Proprietary |
| 271 | GPT-4 (0314)OpenAI | 1284.00 | +/-10 | 7,052 | OpenAI | Proprietary |
| 272 | Mistral Large (2411)Mistral | 1282.00 | +/-9 | 3,574 | Mistral | MRL |
| 273 | Granite-4.0-H-SmallIBM | 1280.00 | +/-32 | 345 | IBM | Apache 2.0 |
| 274 | Hunyuan Large VisionTencent | 1280.00 | +/-30 | 345 | Tencent | Proprietary |
| 275 | Hunyuan Turbo (0110)Tencent | 1279.00 | +/-31 | 243 | Tencent | Proprietary |
| 276 | Llama 3.1 Nemotron 70B InstructNvidia | 1279.00 | +/-17 | 1,041 | Nvidia | Llama 3.1 |
| 277 | GPT-4 (0613)OpenAI | 1277.00 | +/-8 | 11,181 | OpenAI | Proprietary |
| 278 | Mistral Small 3.1 24B Instruct (2503)Mistral | 1277.00 | +/-13 | 2,105 | Mistral | Apache 2.0 |
| 279 | GPT-4o mini (2024-07-18)OpenAI | 1275.00 | +/-7 | 9,322 | OpenAI | Proprietary |
| 280 | GPT-4.1 nano (2025-04-14)OpenAI | 1273.00 | +/-23 | 582 | OpenAI | Proprietary |
| 281 | Qwen2-72B-Instruct阿里巴巴 | 1273.00 | +/-9 | 4,835 | 阿里巴巴 | Qianwen LICENSE |
| 282 | Grok 2 mini (2024-08-13)SpaceXAI | 1273.00 | +/-8 | 7,261 | SpaceXAI | Proprietary |
| 283 | DeepSeek-Coder-V2DeepSeek | 1272.00 | +/-14 | 1,858 | DeepSeek | DeepSeek License |
| 284 | Llama 3.1 Nemotron 51B InstructNvidia | 1271.00 | +/-22 | 507 | Nvidia | Llama 3.1 |
| 285 | Qwen2.5-Coder-32B-Instruct阿里巴巴 | 1269.00 | +/-19 | 725 | 阿里巴巴 | Apache 2.0 |
| 286 | Llama 3.1 70B InstructMeta | 1269.00 | +/-8 | 7,677 | Meta | Llama 3.1 Community |
| 287 | Amazon Nova Pro 1.0Amazon | 1269.00 | +/-10 | 2,978 | Amazon | Proprietary |
| 288 | Phi-4-reasoning微软 | 1265.00 | +/-11 | 2,764 | 微软 | MIT |
| 289 | Llama 3.1 Tulu 3 70BAi2 | 1263.00 | +/-25 | 397 | Ai2 | Llama 3.1 |
| 290 | Athene 70B (0725) CC-BY-NC-4.0 | 1263.00 | +/-10 | 2,921 | — | — |
| 291 | Mistral Small 24B Instruct (2501)Mistral | 1261.00 | +/-13 | 1,683 | Mistral | Apache 2.0 |
| 292 | Gemma 3n E4B ITGoogle | 1260.00 | +/-15 | 1,562 | Gemma | |
| 293 | Gemini 1.5 Flash (001)Google | 1259.00 | +/-8 | 8,392 | Proprietary | |
| 294 | Llama 3 70B InstructMeta | 1258.00 | +/-8 | 20,941 | Meta | Llama 3 Community |
| 295 | Gemma 3 4B ITGoogle | 1254.00 | +/-28 | 423 | Gemma | |
| 296 | Claude 3 Sonnet (20240229)Anthropic | 1254.00 | +/-8 | 13,766 | Anthropic | Proprietary |
| 297 | Nemotron-4 340B InstructNvidia | 1252.00 | +/-12 | 2,352 | Nvidia | NVIDIA Open Model |
| 298 | Hunyuan Standard 256KTencent | 1249.00 | +/-29 | 361 | Tencent | Proprietary |
| 299 | Gemma 2 27B ITGoogle | 1248.00 | +/-7 | 10,170 | Gemma license | |
| 300 | GLM-4 (0520)Z.ai | 1246.00 | +/-16 | 1,191 | Z.ai | Proprietary |
| 301 | Reka Core (20240904) Proprietary | 1246.00 | +/-14 | 1,207 | — | — |
| 302 | Jamba 1.5 Large Jamba Open | 1245.00 | +/-15 | 1,147 | — | — |
| 303 | Mistral Large (2402)Mistral | 1245.00 | +/-9 | 7,987 | Mistral | Proprietary |
| 304 | Amazon Nova Lite 1.0Amazon | 1244.00 | +/-11 | 2,511 | Amazon | Proprietary |
| 305 | Aya Expanse 32BCohere | 1234.00 | +/-10 | 3,854 | Cohere | CC-BY-NC-4.0 |
| 306 | Reka Flash (20240904) Proprietary | 1233.00 | +/-14 | 1,284 | — | — |
| 307 | Claude 3 Haiku (20240307)Anthropic | 1232.00 | +/-8 | 14,983 | Anthropic | Proprietary |
| 308 | Command R+ (08-2024)Cohere | 1232.00 | +/-14 | 1,467 | Cohere | CC-BY-NC-4.0 |
| 309 | Gemini 1.5 Flash-8B (001)Google | 1230.00 | +/-9 | 5,036 | Proprietary | |
| 310 | Mixtral 8x22B Instruct v0.1Mistral | 1228.00 | +/-9 | 6,778 | Mistral | Apache 2.0 |
| 311 | OLMo 2 32B Instruct (0325)Ai2 | 1227.00 | +/-28 | 375 | Ai2 | Apache-2.0 |
| 312 | Amazon Nova Micro 1.0Amazon | 1223.00 | +/-11 | 2,455 | Amazon | Proprietary |
| 313 | Qwen1.5-110B-Chat阿里巴巴 | 1221.00 | +/-11 | 3,188 | 阿里巴巴 | Qianwen LICENSE |
| 314 | Gemma 2 9B ITGoogle | 1221.00 | +/-8 | 7,110 | Gemma license | |
| 315 | Mistral MediumMistral | 1220.00 | +/-11 | 4,406 | Mistral | Proprietary |
| 316 |  Phi-3 Medium 4K InstructMicrosoft | 1216.00 | +/-11 | 3,238 | Microsoft | MIT |
| 317 | Command R+Cohere | 1214.00 | +/-9 | 9,769 | Cohere | CC-BY-NC-4.0 |
| 318 | Ministral 8B (2410)Mistral | 1214.00 | +/-20 | 683 | Mistral | MRL |
| 319 | Yi 1.5 34B Chat Apache-2.0 | 1212.00 | +/-11 | 2,985 | — | — |
| 320 | Reka Flash 21B (Online, 20240226) Proprietary | 1212.00 | +/-14 | 2,028 | — | — |
| 321 | Qwen1.5-72B-Chat阿里巴巴 | 1209.00 | +/-10 | 5,327 | 阿里巴巴 | Qianwen LICENSE |
| 322 | QwQ-32B-Preview阿里巴巴 | 1209.00 | +/-24 | 480 | 阿里巴巴 | Apache 2.0 |
| 323 | Gemma 2 9B IT SimPO MIT | 1208.00 | +/-15 | 1,285 | — | — |
| 324 | Command R (08-2024)Cohere | 1207.00 | +/-14 | 1,601 | Cohere | CC-BY-NC-4.0 |
| 325 | InternLM2.5 20B Chat Other | 1206.00 | +/-15 | 1,387 | — | — |
| 326 | Llama 3.1 Tulu 3 8BAi2 | 1206.00 | +/-26 | 363 | Ai2 | Llama 3.1 |
| 327 | GPT-3.5 Turbo (1106)OpenAI | 1205.00 | +/-15 | 2,134 | OpenAI | Proprietary |
| 328 | Gemini ProGoogle | 1203.00 | +/-19 | 993 | Proprietary | |
| 329 | GPT-3.5 Turbo (0125)OpenAI | 1202.00 | +/-9 | 8,626 | OpenAI | Proprietary |
| 330 | Aya Expanse 8BCohere | 1201.00 | +/-15 | 1,307 | Cohere | CC-BY-NC-4.0 |
| 331 | Qwen1.5-32B-ChatAlibaba | 1201.00 | +/-12 | 2,649 | Alibaba | Qianwen LICENSE |
| 332 | Reka Flash 21B (20240226) Proprietary | 1199.00 | +/-11 | 3,363 | — | — |
| 333 | Gemini Pro (Dev API)Google | 1199.00 | +/-14 | 2,274 | Proprietary | |
| 334 | Granite 3.0 8B InstructIBM | 1198.00 | +/-19 | 873 | IBM | Apache 2.0 |
| 335 | Granite 3.1 2B InstructIBM | 1198.00 | +/-26 | 391 | IBM | Apache 2.0 |
| 336 | Zephyr ORPO 141B A35B v0.1 Apache 2.0 | 1197.00 | +/-22 | 589 | — | — |
| 337 | DBRX Instruct Preview DBRX LICENSE | 1197.00 | +/-12 | 4,001 | — | — |
| 338 | Phi-3 Mini 4K Instruct (June 2024)Microsoft | 1194.00 | +/-14 | 1,568 | Microsoft | MIT |
| 339 | Phi-3 Small 8K InstructMicrosoft | 1193.00 | +/-13 | 2,092 | Microsoft | MIT |
| 340 | Llama 3 8B InstructMeta | 1193.00 | +/-8 | 14,252 | Meta | Llama 3 Community |
| 341 | Mixtral 8x7B Instruct v0.1Mistral | 1191.00 | +/-9 | 9,663 | Mistral | Apache 2.0 |
| 342 | Granite 3.1 8B InstructIBM | 1190.00 | +/-28 | 382 | IBM | Apache 2.0 |
| 343 | Llama 3.1 8B InstructMeta | 1188.00 | +/-8 | 7,135 | Meta | Llama 3.1 Community |
| 344 | Jamba 1.5 Mini Jamba Open | 1187.00 | +/-16 | 1,094 | — | — |
| 345 | Command RCohere | 1177.00 | +/-10 | 6,682 | Cohere | CC-BY-NC-4.0 |
| 346 | Granite 3.0 2B InstructIBM | 1170.00 | +/-19 | 908 | IBM | Apache 2.0 |
| 347 | Qwen1.5-14B-Chat阿里巴巴 | 1167.00 | +/-14 | 2,184 | 阿里巴巴 | Qianwen LICENSE |
| 348 | Llama 3.2 3B InstructMeta | 1165.00 | +/-16 | 1,136 | Meta | Llama 3.2 |
| 349 | Gemma 2 2B ITGoogle | 1165.00 | +/-8 | 6,599 | Gemma license | |
| 350 | Snowflake Arctic Instruct Apache 2.0 | 1163.00 | +/-11 | 4,793 | — | — |
| 351 | Gemma 1.1 7B ITGoogle | 1163.00 | +/-11 | 3,039 | Gemma license | |
| 352 | OpenChat 3.5 (0106) Apache-2.0 | 1159.00 | +/-14 | 1,726 | — | — |
| 353 | WizardLM 70BMicrosoft | 1158.00 | +/-19 | 903 | Microsoft | Llama 2 Community |
| 354 | Starling LM 7B Beta Apache-2.0 | 1157.00 | +/-14 | 1,973 | — | — |
| 355 | DeepSeek LLM 67B ChatDeepSeek | 1156.00 | +/-24 | 576 | DeepSeek | DeepSeek License |
| 356 | SmolLM2 1.7B Instruct Apache 2.0 | 1153.00 | +/-33 | 271 | — | — |
| 357 | OpenHermes 2.5 Mistral 7B Apache-2.0 | 1153.00 | +/-20 | 697 | — | — |
| 358 | Yi 34B Chat Yi License | 1151.00 | +/-13 | 2,043 | — | — |
| 359 | Phi-3 Mini 4K InstructMicrosoft | 1150.00 | +/-12 | 2,564 | Microsoft | MIT |
| 360 | Tulu 2 DPO 70B AI2 ImpACT Low-risk | 1144.00 | +/-19 | 888 | — | — |
| 361 | Phi-3 Mini 128K InstructMicrosoft | 1140.00 | +/-13 | 2,813 | Microsoft | MIT |
| 362 | Llama 2 70B ChatMeta | 1137.00 | +/-10 | 4,740 | Meta | Llama 2 Community |
| 363 | Mistral 7B Instruct v0.2Mistral | 1127.00 | +/-12 | 2,605 | Mistral | Apache-2.0 |
| 364 | Qwen-14B-Chat阿里巴巴 | 1127.00 | +/-24 | 534 | 阿里巴巴 | Qianwen LICENSE |
| 365 | Starling LM 7B Alpha CC-BY-NC-4.0 | 1127.00 | +/-16 | 1,300 | — | — |
| 366 | OpenChat 3.5 Apache-2.0 | 1126.00 | +/-18 | 945 | — | — |
| 367 | Dolphin 2.2.1 Mistral 7B Apache-2.0 | 1125.00 | +/-32 | 219 | — | — |
| 368 | Llama 3.2 1B InstructMeta | 1123.00 | +/-16 | 1,162 | Meta | Llama 3.2 |
| 369 | Qwen1.5-7B-Chat阿里巴巴 | 1121.00 | +/-21 | 690 | 阿里巴巴 | Qianwen LICENSE |
| 370 | Gemma 7B ITGoogle | 1120.00 | +/-17 | 1,120 | Gemma license | |
| 371 | PaLM 2Google | 1117.00 | +/-19 | 901 | Proprietary | |
| 372 | Vicuna 33B Non-commercial | 1116.00 | +/-13 | 2,663 | — | — |
| 373 | Llama 2 70B SteerLM ChatNvidia | 1114.00 | +/-27 | 440 | Nvidia | Llama 2 Community |
| 374 | Gemma 1.1 2B ITGoogle | 1111.00 | +/-16 | 1,355 | Gemma license | |
| 375 | Llama 2 13B ChatMeta | 1110.00 | +/-13 | 2,218 | Meta | Llama 2 Community |
| 376 | Code Llama 34B InstructMeta | 1110.00 | +/-19 | 770 | Meta | Llama 2 Community |
| 377 | SOLAR 10.7B Instruct v1.0Upstage | 1109.00 | +/-22 | 604 | Upstage | CC-BY-NC-4.0 |
| 378 | MPT 30B Chat CC-BY-NC-SA-4.0 | 1096.00 | +/-34 | 242 | — | — |
| 379 | Nous Hermes 2 Mixtral 8x7B DPO Apache-2.0 | 1093.00 | +/-21 | 628 | — | — |
| 380 | Qwen1.5-4B-Chat阿里巴巴 | 1088.00 | +/-18 | 988 | 阿里巴巴 | Qianwen LICENSE |
| 381 | Llama 2 7B ChatMeta | 1086.00 | +/-14 | 1,656 | Meta | Llama 2 Community |
| 382 | StripedHyena Nous 7B Apache 2.0 | 1085.00 | +/-20 | 676 | — | — |
| 383 | Vicuna 13B Llama 2 Community | 1083.00 | +/-14 | 2,146 | — | — |
| 384 | Mistral 7B InstructMistral | 1083.00 | +/-19 | 974 | Mistral | Apache 2.0 |
| 385 | Zephyr 7B Beta MIT | 1082.00 | +/-17 | 1,250 | — | — |
| 386 | Guanaco 33B Non-commercial | 1081.00 | +/-32 | 280 | — | — |
| 387 | Gemma 2B ITGoogle | 1074.00 | +/-22 | 597 | Gemma license | |
| 388 | WizardLM 13BMicrosoft | 1065.00 | +/-21 | 669 | Microsoft | Llama 2 Community |
| 389 | OLMo 7B InstructAi2 | 1053.00 | +/-19 | 848 | Ai2 | Apache-2.0 |
| 390 | Vicuna 7B Llama 2 Community | 1048.00 | +/-22 | 658 | — | — |
| 391 | ChatGLM3 6B Apache-2.0 | 1043.00 | +/-23 | 576 | — | — |
| 392 | GPT4All 13B Snoozy Non-commercial | 999.00 | +/-37 | 211 | — | — |
| 393 | Alpaca 13B Non-commercial | 996.00 | +/-23 | 652 | — | — |
| 394 | MPT 7B Chat CC-BY-NC-SA-4.0 | 987.00 | +/-25 | 471 | — | — |
| 395 | RWKV-4 Raven 14B Apache 2.0 | 985.00 | +/-24 | 544 | — | — |
| 396 | Koala 13B Non-commercial | 980.00 | +/-21 | 751 | — | — |
| 397 | ChatGLM 6B Non-commercial | 977.00 | +/-26 | 525 | — | — |
| 398 | ChatGLM2 6B Apache-2.0 | 973.00 | +/-35 | 227 | — | — |
| 399 | OpenAssistant Pythia 12B Apache 2.0 | 962.00 | +/-22 | 687 | — | — |
| 400 | Dolly 2.0 12B MIT | 954.00 | +/-29 | 370 | — | — |
| 401 | LLaMA 13BMeta | 923.00 | +/-33 | 252 | Meta | Non-commercial |
| 402 | FastChat-T5 3B Apache 2.0 | 921.00 | +/-26 | 462 | — | — |
| 403 | StableLM Tuned Alpha 7B CC-BY-NC-SA-4.0 | 891.00 | +/-29 | 353 | — | — |
数据仅供参考，以官方来源为准。模型名称旁的链接可跳转到 DataLearner 模型详情页。
常见问题 (FAQ)
01
什么是 LMArena Math Arena？
LMArena Math Arena 是 LMArena 旗下专注于数学推理能力的匿名评测平台。用户提交真实数学问题（如代数、几何、竞赛数学等），系统将不同模型的解题过程并排展示（隐藏模型名称），由用户投票选出更好的解答，最终通过 Elo 算法汇总形成动态排行榜。
02
Math Arena 与 MATH-500、AIME 等静态基准有什么区别？
MATH-500、AIME、AMC 等静态基准使用固定题目集和自动评分，可重现性强但容易被针对性优化（"刷榜"）。Math Arena 来自真实用户的开放式数学问题，测试内容不固定，更能反映模型在实际数学场景中的自然表现，两者互为补充。
03
思考模型（Thinking Model）在数学 Arena 中表现更好吗？
整体而言，具备思维链（Chain-of-Thought）或扩展推理能力的模型在数学 Arena 中往往排名更高。Claude Opus 系列 Thinking 模式、GPT 高算力模式以及 DeepSeek 思考版本均在榜单前列，说明延长推理时间对数学问题的解答质量有显著提升。
04
国产大模型在数学能力方面表现如何？
DeepSeek、Qwen3 系列、GLM 等国产模型在 Math Arena 表现亮眼，已跻身全球前列。DeepSeek 以 MIT 协议开源，Qwen3-235B 等系列支持中文数学场景，是选择开源数学推理模型的重要参考。