Text Generation Arena 文本生成模型排行榜
基于 Text Generation Arena 用户匿名投票的最新AI文本生成模型排行榜，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
GPT-5.5 (xhigh)
最高得分
60
模型数量
200
数据版本
2026年04月24日
数据来源: LM Arena
关于本排行榜
本排行榜展示了当前最强 AI 大模型在文本生成任务中的综合实力排名。数据来源于 LMArena（前身为 LMSYS Chatbot Arena），这是目前全球最大的 AI 模型众包评测平台。用户在平台上与两个匿名模型同时对话，并投票选出更好的回答——排名完全由真实用户的偏好决定，而非实验室基准测试。
评测方法概要
匿名盲测：用户同时与两个"隐藏身份"的模型对话，根据回答质量投票，排除品牌偏见。
Elo 评分：基于国际象棋领域的 Elo Rating 体系（Bradley-Terry 模型），通过对战结果计算每个模型的实力分数。分数越高，说明模型在真实对话中被用户选中的概率越大。
场景覆盖广泛：涵盖编程、创意写作、数学推理、知识问答、角色扮演等高频真实场景。
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。
筛选条件
榜单历史快照月份:
排名总表
| 排名 | 模型名称 | 得分 | 95% CI | 投票数 | 机构 | 许可证 |
|---|---|---|---|---|---|---|
| 1 | GPT-5.5 (xhigh) | 60 | / | / | OpenAI | / |
| 2 | GPT-5.5 (high) | 59 | / | / | OpenAI | / |
| 3 | Claude Opus 4.7 (max) | 57 | / | / | Anthropic | / |
| 4 | Gemini 3.1 Pro Preview | 57 | / | / | Google Deep Mind | / |
| 5 | GPT-5.4 | 57 | / | / | OpenAI | / |
| 6 | GPT-5.5 (medium) | 57 | / | / | OpenAI | / |
| 7 | Kimi K2.6 | 54 | / | / | Kimi | / |
| 8 | MiMo-V2.5-Pro | 54 | / | / | Xiaomi | / |
| 9 | GPT-5.3 Codex | 54 | / | / | OpenAI | / |
| 10 | Muse Spark | 52 | / | / | Facebook AI研究实验室 | / |
| 11 | Claude Opus 4.7 (Non-reasoning, high) | 52 | / | / | Anthropic | / |
| 12 | Qwen3.6 Max Preview | 52 | / | / | Alibaba | / |
| 13 | Claude Sonnet 4.6 | 52 | / | / | Anthropic | / |
| 14 | GLM 5.1 | 51 | / | / | 智谱AI | / |
| 15 | GPT-5.5 (low) | 51 | / | / | OpenAI | / |
| 16 | Qwen 3.6 Plus Preview | 50 | / | / | 阿里巴巴 | / |
| 17 | GLM-5 | 50 | / | / | 智谱AI | / |
| 18 | MiniMax-M2.7 | 50 | / | / | MiniMax | / |
| 19 | Grok 4.20 0309 v2 | 49 | / | / | xAI | / |
| 20 | MiMo-V2-Pro | 49 | / | / | Xiaomi | / |
| 21 | GPT-5.4 mini | 49 | / | / | OpenAI | / |
| 22 | GLM-5-Turbo | 47 | / | / | 智谱AI | / |
| 23 | DeepSeek V4 Flash (Max) | 47 | / | / | DeepSeek | / |
| 24 | Gemini 3.0 Flash | 46 | / | / | Google Deep Mind | / |
| 25 | Qwen3.6 27B | 46 | / | / | Alibaba | / |
| 26 | Qwen3.5-397B-A17B | 45 | / | / | 阿里巴巴 | / |
| 27 | Nova 2 Omni（Preview） | 45 | / | / | 亚马逊 | / |
| 28 | DeepSeek V4 Flash (High) | 45 | / | / | DeepSeek | / |
| 29 | Claude Sonnet 4.6 | 44 | / | / | Anthropic | / |
| 30 | GPT-5.4 nano | 44 | / | / | OpenAI | / |
| 31 | GLM 5.1 | 44 | / | / | 智谱AI | / |
| 32 | Qwen3.6 35B A3B | 43 | / | / | Alibaba | / |
| 33 | MiMo-V2-Omni | 43 | / | / | Xiaomi | / |
| 34 | GLM-5V-Turbo | 43 | / | / | 智谱AI | / |
| 35 | Claude Sonnet 4.6 | 43 | / | / | Anthropic | / |
| 36 | DeepSeek V3.2 | 42 | / | / | DeepSeek-AI | / |
| 37 | Qwen3.5-122B-A10B | 42 | / | / | 阿里巴巴 | / |
| 38 | Gemini 2.0 Flash Experimental | 41 | / | / | DeepMind | / |
| 39 | Gemini 3.1 Pro Preview | 41 | / | / | Google Deep Mind | / |
| 40 | GPT-5.5 (Non-reasoning) | 41 | / | / | OpenAI | / |
| 41 | GLM-5 | 41 | / | / | 智谱AI | / |
| 42 | Qwen3.5-397B-A17B | 40 | / | / | 阿里巴巴 | / |
| 43 | Gemma 4 31B | 39 | / | / | / | |
| 44 | Qwen3.5-Omni-Plus | 39 | / | / | 阿里巴巴 | / |
| 45 | Grok 4.1 Fast | 39 | / | / | xAI | / |
| 46 | Step 3.5 Flash | 38 | / | / | StepFunAI | / |
| 47 | OpenAI o3 | 38 | / | / | OpenAI | / |
| 48 | GPT-5.4 nano | 38 | / | / | OpenAI | / |
| 49 | GPT-5.4 mini | 38 | / | / | OpenAI | / |
| 50 | Kimi K2.5 | 37 | / | / | Moonshot AI | / |
| 51 | Haiku 4.5 | 37 | / | / | Anthropic | / |
| 52 | NVIDIA Nemotron 3 Super | 36 | / | / | NVIDIA | / |
| 53 | Qwen3.5-122B-A10B | 36 | / | / | 阿里巴巴 | / |
| 54 | Nova 2 Pro（Preview） | 36 | / | / | 亚马逊 | / |
| 55 | GPT-5.4 (Non-reasoning) | 35 | / | / | OpenAI | / |
| 56 | Gemini 3.0 Flash | 35 | / | / | Google Deep Mind | / |
| 57 | Gemini 2.5-Pro | 35 | / | / | Google Deep Mind | / |
| 58 | Nova 2 Lite | 35 | / | / | 亚马逊 | / |
| 59 | Ling-2.6-1T | 34 | / | / | InclusionAI | / |
| 60 | Gemini 3.1 Flash-Lite Preview | 34 | / | / | / | |
| 61 | Doubao Seed Code | 34 | / | / | ByteDance Seed | / |
| 62 | GPT OSS 120B | 33 | / | / | OpenAI | / |
| 63 | Mercury 2 | 33 | / | / | Inception | / |
| 64 | Qwen3.5-9B-Instruct | 32 | / | / | 阿里巴巴 | / |
| 65 | Gemma 4 31B | 32 | / | / | / | |
| 66 | K-EXAONE | 32 | / | / | LG AI Research | / |
| 67 | DeepSeek V3.2 | 32 | / | / | DeepSeek-AI | / |
| 68 | Grok-3 mini - Reasoning | 32 | / | / | xAI | / |
| 69 | Nova 2 Pro（Preview） | 32 | / | / | 亚马逊 | / |
| 70 | Trinity Large Thinking | 32 | / | / | Arcee AI | / |
| 71 | Qwen3.6 35B A3B | 32 | / | / | Alibaba | / |
| 72 | Gemma 4 26B A4B | 31 | / | / | / | |
| 73 | Haiku 4.5 | 31 | / | / | Anthropic | / |
| 74 | Qwen3.5-35B-A3B | 31 | / | / | 阿里巴巴 | / |
| 75 | MiMo-V2-Flash | 30 | / | / | Xiaomi | / |
| 76 | Nova 2 Lite | 30 | / | / | 亚马逊 | / |
| 77 | DeepSeek V3.2 Speciale | 29 | / | / | DeepSeek-AI | / |
| 78 | ERNIE 5.0 | 29 | / | / | 百度 | / |
| 79 | Grok 4.20 0309 v2 | 29 | / | / | xAI | / |
| 80 | Grok Code Fast 1 | 29 | / | / | xAI | / |
| 81 | Nemotron Cascade 2 30B A3B | 28 | / | / | NVIDIA | / |
| 82 | Qwen3-Coder-Next | 28 | / | / | 阿里巴巴 | / |
| 83 | Nova 2 Omni（Preview） | 28 | / | / | 亚马逊 | / |
| 84 | Mistral Small 4 | 28 | / | / | Mistral | / |
| 85 | Qwen3.5-9B-Instruct | 27 | / | / | 阿里巴巴 | / |
| 86 | Magistral Medium 1.2 | 27 | / | / | Mistral | / |
| 87 | Gemma 4 26B A4B | 27 | / | / | / | |
| 88 | Qwen3.5 4B | 27 | / | / | Alibaba | / |
| 89 | DeepSeek-R1-0528 | 27 | / | / | DeepSeek-AI | / |
| 90 | Qwen3-Next | 27 | / | / | 阿里巴巴 | / |
| 91 | Ling 2.6 Flash | 26 | / | / | InclusionAI | / |
| 92 | Solar Pro 3 | 26 | / | / | Upstage | / |
| 93 | Qwen3.5-Omni-Flash | 26 | / | / | 阿里巴巴 | / |
| 94 | JT-MINI | 25 | / | / | China Mobile | / |
| 95 | Nova 2 Lite | 25 | / | / | 亚马逊 | / |
| 96 | GPT OSS 20B | 24 | / | / | OpenAI | / |
| 97 | GPT OSS 120B | 24 | / | / | OpenAI | / |
| 98 | GPT-5.4 nano | 24 | / | / | OpenAI | / |
| 99 | NVIDIA Nemotron 3 Nano | 24 | / | / | NVIDIA | / |
| 100 | LongCat Flash Lite | 24 | / | / | LongCat | / |
| 101 | Grok 4.1 Fast | 24 | / | / | xAI | / |
| 102 | K-EXAONE | 23 | / | / | LG AI Research | / |
| 103 | GPT-5.4 mini | 23 | / | / | OpenAI | / |
| 104 | Nova 2 Omni（Preview） | 23 | / | / | 亚马逊 | / |
| 105 | Nova 2 Pro（Preview） | 23 | / | / | 亚马逊 | / |
| 106 | Mi:dm K 2.5 Pro | 23 | / | / | Korea Telecom | / |
| 107 | Mistral Large 3 | 23 | / | / | MistralAI | / |
| 108 | Ring-1T | 23 | / | / | InclusionAI | / |
| 109 | Qwen3.5 4B | 23 | / | / | Alibaba | / |
| 110 | INTELLECT-3 | 22 | / | / | Prime Intellect | / |
| 111 | Devstral 2 | 22 | / | / | Mistral | / |
| 112 | Solar Open 100B | 22 | / | / | Upstage | / |
| 113 | Gemini 2.5 Flash-Lite-Preview-09-2025 | 22 | / | / | Google Deep Mind | / |
| 114 | Mistral Medium 3.1 | 21 | / | / | Mistral | / |
| 115 | GPT OSS 20B | 21 | / | / | OpenAI | / |
| 116 | Qwen3-Next | 20 | / | / | 阿里巴巴 | / |
| 117 | Devstral Small 2 | 19 | / | / | Mistral | / |
| 118 | Gemini 2.5 Flash-Lite-Preview-09-2025 | 19 | / | / | Google Deep Mind | / |
| 119 | Motif-2-12.7B | 19 | / | / | Motif Technologies | / |
| 120 | Ling-1T | 19 | / | / | InclusionAI | / |
| 121 | Nova Premier | 19 | / | / | Amazon | / |
| 122 | Gemma 4 E4B | 19 | / | / | / | |
| 123 | Llama Nemotron Super 49B v1.5 | 19 | / | / | NVIDIA | / |
| 124 | Mistral Small 4 | 19 | / | / | Mistral | / |
| 125 | Llama 3.3 Nemotron Super 49B | 18 | / | / | NVIDIA | / |
| 126 | Llama 4 Maverick | 18 | / | / | Facebook AI研究实验室 | / |
| 127 | Magistral Small 1.2 | 18 | / | / | Mistral | / |
| 128 | Sarvam 105B (high) | 18 | / | / | Sarvam | / |
| 129 | Nova 2 Lite | 18 | / | / | 亚马逊 | / |
| 130 | Llama3.1-405B | 17 | / | / | Facebook AI研究实验室 | / |
| 131 | EXAONE 4.0 32B | 17 | / | / | LG AI Research | / |
| 132 | Nova 2 Omni（Preview） | 17 | / | / | 亚马逊 | / |
| 133 | Qwen3.5 2B | 16 | / | / | Alibaba | / |
| 134 | Nanbeige4.1-3B | 16 | / | / | Nanbeige | / |
| 135 | Ministral 3 14B | 16 | / | / | MistralAI | / |
| 136 | DeepSeek-R1-Distill-Llama-70B | 16 | / | / | DeepSeek-AI | / |
| 137 | Falcon-H1R-7B | 16 | / | / | TII UAE | / |
| 138 | Ling-flash-2.0 | 16 | / | / | InclusionAI | / |
| 139 | Qwen3-Omni-30B-A3B | 16 | / | / | 阿里巴巴 | / |
| 140 | Step3 VL 10B | 15 | / | / | StepFun | / |
| 141 | Gemma 4 E2B | 15 | / | / | / | |
| 142 | Llama Nemotron Ultra | 15 | / | / | NVIDIA | / |
| 143 | ERNIE-4.5-300B-A47B | 15 | / | / | 百度 | / |
| 144 | Solar Pro 2 | 15 | / | / | Upstage | / |
| 145 | NVIDIA Nemotron Nano 12B v2 VL | 15 | / | / | NVIDIA | / |
| 146 | Ministral 3 8B | 15 | / | / | MistralAI | / |
| 147 | Gemma 4 E4B | 15 | / | / | / | |
| 148 | NVIDIA Nemotron Nano 9B V2 | 15 | / | / | NVIDIA | / |
| 149 | NVIDIA Nemotron 3 Nano 4B | 15 | / | / | NVIDIA | / |
| 150 | Qwen3.5 2B | 15 | / | / | Alibaba | / |
| 151 | Llama Nemotron Super 49B v1.5 | 15 | / | / | NVIDIA | / |
| 152 | Llama3.3-70B-Instruct | 14 | / | / | Facebook AI研究实验室 | / |
| 153 | Llama 3.1 Nemotron Nano 4B v1.1 | 14 | / | / | NVIDIA | / |
| 154 | Kimi Linear 48B A3B Instruct | 14 | / | / | Kimi | / |
| 155 | Llama 3.3 Nemotron Super 49B | 14 | / | / | NVIDIA | / |
| 156 | Ring-flash-2.0 | 14 | / | / | InclusionAI | / |
| 157 | Solar Pro 2 | 14 | / | / | Upstage | / |
| 158 | Llama 4 Scout | 14 | / | / | Facebook AI研究实验室 | / |
| 159 | C4AI Command A (202503) | 13 | / | / | CohereAI | / |
| 160 | Llama 3.1 Nemotron 70B | 13 | / | / | NVIDIA | / |
| 161 | NVIDIA Nemotron 3 Nano | 13 | / | / | NVIDIA | / |
| 162 | NVIDIA Nemotron Nano 9B V2 | 13 | / | / | NVIDIA | / |
| 163 | Sarvam 30B (high) | 12 | / | / | Sarvam | / |
| 164 | Gemma 4 E2B | 12 | / | / | / | |
| 165 | R1 1776 | 12 | / | / | Perplexity | / |
| 166 | Llama 3.2-Vision-90B | 12 | / | / | Facebook AI研究实验室 | / |
| 167 | EXAONE 4.0 32B | 12 | / | / | LG AI Research | / |
| 168 | Ministral 3 3B | 11 | / | / | Mistral | / |
| 169 | Jamba 1.7 Large | 11 | / | / | AI21 Labs | / |
| 170 | Granite 4.0 H Small | 11 | / | / | IBM | / |
| 171 | Qwen3-Omni-30B-A3B | 11 | / | / | 阿里巴巴 | / |
| 172 | Qwen3.5 0.8B | 11 | / | / | Alibaba | / |
| 173 | LFM2 24B A2B | 10 | / | / | Liquid AI | / |
| 174 | Phi 4 - 14B | 10 | / | / | Microsoft Azure | / |
| 175 | Amazon Nova Micro | 10 | / | / | 亚马逊 | / |
| 176 | NVIDIA Nemotron Nano 12B v2 VL | 10 | / | / | NVIDIA | / |
| 177 | Phi-4-multimodal-instruct | 10 | / | / | Microsoft Azure | / |
| 178 | Qwen3.5 0.8B | 10 | / | / | Alibaba | / |
| 179 | Jamba Reasoning 3B | 10 | / | / | AI21 Labs | / |
| 180 | Gemini 3.0 Flash | 10 | / | / | Google Deep Mind | / |
| 181 | Ling-mini-2.0 | 9 | / | / | InclusionAI | / |
| 182 | Llama 3.2-Vision-11B | 9 | / | / | Facebook AI研究实验室 | / |
| 183 | Phi-4-mini-instruct (3.8B) | 8 | / | / | Microsoft Azure | / |
| 184 | Exaone 4.0 1.2B | 8 | / | / | LG AI Research | / |
| 185 | Exaone 4.0 1.2B | 8 | / | / | LG AI Research | / |
| 186 | LFM2.5-1.2B-Thinking | 8 | / | / | Liquid AI | / |
| 187 | Jamba 1.7 Mini | 8 | / | / | AI21 Labs | / |
| 188 | LFM2.5-1.2B-Instruct | 8 | / | / | Liquid AI | / |
| 189 | LFM2 2.6B | 8 | / | / | Liquid AI | / |
| 190 | Granite 4.0 H 1B | 8 | / | / | IBM | / |
| 191 | Gemma 3-270M | 8 | / | / | Google Deep Mind | / |
| 192 | Apertus 70B Instruct | 8 | / | / | Swiss AI Initiative | / |
| 193 | Granite 4.0 Micro | 8 | / | / | IBM | / |
| 194 | Granite 4.0 1B | 7 | / | / | IBM | / |
| 195 | LFM2 8B A1B | 7 | / | / | Liquid AI | / |
| 196 | LFM2.5-VL-1.6B | 6 | / | / | Liquid AI | / |
| 197 | Granite 4.0 350M | 6 | / | / | IBM | / |
| 198 | Apertus 8B Instruct | 6 | / | / | Swiss AI Initiative | / |
| 199 | Granite 4.0 H 350M | 5 | / | / | IBM | / |
| 200 | Tiny Aya Global | 5 | / | / | Cohere | / |
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。