加载中...
加载中...
Artificial Analysis Intelligence Index v4.0 综合了10项权威评测基准（GDPval-AA、Terminal-Bench、GPQA Diamond、SciCode等），从数学、科学、编程、推理等多维度对AI模型进行全面评估和排名。
榜首模型
Gemini 3.1 Pro Preview
最高得分
57
模型数量
196
数据版本
2026年03月26日
数据来源: Artificial Analysis
图表来源：DataLearnerAI · 数据来源：LMArena
| 排名 | 模型名称 | 智能指数 | 机构 |
|---|---|---|---|
| 1 | Gemini 3.1 Pro Preview | 57 | |
| 2 | GPT-5.4 (xhigh) | 57 | OpenAI |
| 3 | GPT-5.3 Codex (xhigh) | 54 | OpenAI |
| 4 | Claude Opus 4.6 (max) | 53 | Anthropic |
| 5 | Claude Sonnet 4.6 (max) | 52 | Anthropic |
| 6 | GLM-5 | 50 | Z AI |
| 7 | MiniMax-M2.7 | 50 | MiniMax |
| 8 | MiMo-V2-Pro | 49 | Xiaomi |
| 9 | Grok 4.20 Beta 0309 | 48 | xAI |
| 10 | GPT-5.4 mini (xhigh) | 48 | OpenAI |
| 11 | Kimi K2.5 | 47 | Kimi |
| 12 | GLM-5-Turbo | 47 | Z AI |
| 13 | Claude Opus 4.6 | 46 | Anthropic |
| 14 | Gemini 3 Flash | 46 | |
| 15 | Qwen3.5 397B A17B | 45 | Alibaba |
| 16 | GPT-5.4 nano (xhigh) | 44 | OpenAI |
| 17 | Claude Sonnet 4.6 | 44 | Anthropic |
| 18 | MiMo-V2-Omni | 43 | Xiaomi |
| 19 | Claude Sonnet 4.6 (Non-reasoning) | 43 | Anthropic |
| 20 | Qwen3.5 27B | 42 | Alibaba |
| 21 | DeepSeek V3.2 | 42 | DeepSeek |
| 22 | Qwen3.5 122B A10B | 42 | Alibaba |
| 23 | MiMo-V2-Flash (Feb 2026) | 41 | Xiaomi |
| 24 | Gemini 3 Pro Preview (low) | 41 | |
| 25 | GLM-5 | 41 | Z AI |
| 26 | Qwen3.5 397B A17B | 40 | Alibaba |
| 27 | Qwen3 Max Thinking | 40 | Alibaba |
| 28 | o3 | 38 | OpenAI |
| 29 | GPT-5.4 nano | 38 | OpenAI |
| 30 | Step 3.5 Flash | 38 | StepFun |
| 31 | GPT-5.4 mini (medium) | 38 | OpenAI |
| 32 | Kimi K2.5 | 37 | Kimi |
| 33 | Qwen3.5 27B | 37 | Alibaba |
| 34 | Qwen3.5 35B A3B | 37 | Alibaba |
| 35 | Claude 4.5 Haiku | 37 | Anthropic |
| 36 | KAT-Coder-Pro V1 | 36 | KwaiKAT |
| 37 | NVIDIA Nemotron 3 Super | 36 | NVIDIA |
| 38 | Qwen3.5 122B A10B | 36 | Alibaba |
| 39 | Nova 2.0 Pro Preview (medium) | 36 | Amazon |
| 40 | GPT-5.4 | 35 | OpenAI |
| 41 | Gemini 3 Flash | 35 | |
| 42 | Gemini 2.5 Pro | 35 | |
| 43 | Gemini 3.1 Flash-Lite Preview | 34 | |
| 44 | Doubao Seed Code | 34 | ByteDance Seed |
| 45 | gpt-oss-120B (high) | 33 | OpenAI |
| 46 | Mercury 2 | 33 | Inception |
| 47 | Qwen3.5 9B | 32 | Alibaba |
| 48 | K-EXAONE | 32 | LG AI Research |
| 49 | DeepSeek V3.2 | 32 | DeepSeek |
| 50 | Grok 3 mini Reasoning (high) | 32 | xAI |
| 51 | Nova 2.0 Pro Preview (low) | 32 | Amazon |
| 52 | Claude 4.5 Haiku | 31 | Anthropic |
| 53 | Qwen3.5 35B A3B | 31 | Alibaba |
| 54 | MiMo-V2-Flash | 30 | Xiaomi |
| 55 | Nova 2.0 Lite (medium) | 30 | Amazon |
| 56 | Grok 4.20 Beta 0309 | 30 | xAI |
| 57 | DeepSeek V3.2 Speciale | 29 | DeepSeek |
| 58 | ERNIE 5.0 Thinking Preview | 29 | Baidu |
| 59 | Grok Code Fast 1 | 29 | xAI |
| 60 | Qwen3 Coder Next | 28 | Alibaba |
| 61 | Nova 2.0 Omni (medium) | 28 | Amazon |
| 62 | Apriel-v1.6-15B-Thinker | 28 | ServiceNow |
| 63 | Qwen3.5 9B | 27 | Alibaba |
| 64 | Magistral Medium 1.2 | 27 | Mistral |
| 65 | Qwen3.5 4B | 27 | Alibaba |
| 66 | DeepSeek R1 0528 | 27 | DeepSeek |
| 67 | Mistral Small 4 | 27 | Mistral |
| 68 | Qwen3 Next 80B A3B | 27 | Alibaba |
| 69 | Qwen3 Coder 480B | 25 | Alibaba |
| 70 | Nova 2.0 Lite (low) | 25 | Amazon |
| 71 | gpt-oss-120B (low) | 24 | OpenAI |
| 72 | gpt-oss-20B (high) | 24 | OpenAI |
| 73 | GPT-5.4 nano | 24 | OpenAI |
| 74 | NVIDIA Nemotron 3 Nano | 24 | NVIDIA |
| 75 | K2 Think V2 | 24 | MBZUAI |
| 76 | LongCat Flash Lite | 24 | LongCat |
| 77 | HyperCLOVA X SEED Think | 24 | Naver |
| 78 | GLM-4.6V | 23 | Z AI |
| 79 | K-EXAONE | 23 | LG AI Research |
| 80 | GPT-5.4 mini | 23 | OpenAI |
| 81 | Nova 2.0 Omni (low) | 23 | Amazon |
| 82 | Nova 2.0 Pro Preview | 23 | Amazon |
| 83 | Mi:dm K 2.5 Pro | 23 | Korea Telecom |
| 84 | Mistral Large 3 | 23 | Mistral |
| 85 | Ring-1T | 23 | InclusionAI |
| 86 | Qwen3.5 4B | 23 | Alibaba |
| 87 | INTELLECT-3 | 22 | Prime Intellect |
| 88 | Devstral 2 | 22 | Mistral |
| 89 | Solar Open 100B | 22 | Upstage |
| 90 | Gemini 2.5 Flash-Lite (Sep) | 22 | |
| 91 | Mistral Medium 3.1 | 21 | Mistral |
| 92 | gpt-oss-20B (low) | 21 | OpenAI |
| 93 | K2-V2 (high) | 21 | MBZUAI |
| 94 | Qwen3 Next 80B A3B | 20 | Alibaba |
| 95 | Tri-21B-think Preview | 20 | Trillion Labs |
| 96 | Devstral Small 2 | 19 | Mistral |
| 97 | Gemini 2.5 Flash-Lite (Sep) | 19 | |
| 98 | Motif-2-12.7B | 19 | Motif Technologies |
| 99 | Ling-1T | 19 | InclusionAI |
| 100 | Nova Premier | 19 | Amazon |
| 101 | Llama Nemotron Super 49B | 19 | NVIDIA |
| 102 | K2-V2 (medium) | 19 | MBZUAI |
| 103 | Mistral Small 4 | 19 | Mistral |
| 104 | Tri-21B-Think | 19 | Trillion Labs |
| 105 | Hermes 4 405B | 19 | Nous Research |
| 106 | Llama 3.3 Nemotron Super | 18 | NVIDIA |
| 107 | Llama 4 Maverick | 18 | Meta |
| 108 | Magistral Small 1.2 | 18 | Mistral |
| 109 | Sarvam 105B (high) | 18 | Sarvam |
| 110 | Nova 2.0 Lite | 18 | Amazon |
| 111 | Hermes 4 405B | 18 | Nous Research |
| 112 | Llama 3.1 405B | 17 | Meta |
| 113 | GLM-4.6V | 17 | Z AI |
| 114 | EXAONE 4.0 32B | 17 | LG AI Research |
| 115 | Nova 2.0 Omni | 17 | Amazon |
| 116 | DeepSeek R1 0528 Qwen3 8B | 16 | DeepSeek |
| 117 | Qwen3.5 2B | 16 | Alibaba |
| 118 | Nanbeige4.1-3B | 16 | Nanbeige |
| 119 | Hermes 4 70B | 16 | Nous Research |
| 120 | Ministral 3 14B | 16 | Mistral |
| 121 | DeepSeek R1 Distill L70B | 16 | DeepSeek |
| 122 | Falcon-H1R-7B | 16 | TII UAE |
| 123 | Ling-flash-2.0 | 16 | InclusionAI |
| 124 | Qwen3 Omni 30B A3B | 16 | Alibaba |
| 125 | Step3 VL 10B | 15 | StepFun |
| 126 | Llama Nemotron Ultra | 15 | NVIDIA |
| 127 | ERNIE 4.5 300B A47B | 15 | Baidu |
| 128 | Solar Pro 2 | 15 | Upstage |
| 129 | NVIDIA Nemotron Nano 12B | 15 | NVIDIA |
| 130 | Ministral 3 8B | 15 | Mistral |
| 131 | NVIDIA Nemotron Nano 9B | 15 | NVIDIA |
| 132 | NVIDIA Nemotron 3 Nano 4B | 15 | NVIDIA |
| 133 | Qwen3.5 2B | 15 | Alibaba |
| 134 | Llama Nemotron Super 49B | 15 | NVIDIA |
| 135 | Llama 3.3 70B | 14 | Meta |
| 136 | K2-V2 (low) | 14 | MBZUAI |
| 137 | Llama 3.1 Nemotron Nano 4B | 14 | NVIDIA |
| 138 | Kimi Linear 48B A3B | 14 | Kimi |
| 139 | Llama 3.3 Nemotron Super | 14 | NVIDIA |
| 140 | Ring-flash-2.0 | 14 | InclusionAI |
| 141 | Olmo 3.1 32B Think | 14 | AI2 |
| 142 | Solar Pro 2 | 14 | Upstage |
| 143 | Llama 4 Scout | 14 | Meta |
| 144 | Command A | 13 | Cohere |
| 145 | Llama 3.1 Nemotron 70B | 13 | NVIDIA |
| 146 | NVIDIA Nemotron 3 Nano | 13 | NVIDIA |
| 147 | NVIDIA Nemotron Nano 9B | 13 | NVIDIA |
| 148 | Hermes 4 70B | 13 | Nous Research |
| 149 | Sarvam 30B (high) | 12 | Sarvam |
| 150 | Olmo 3.1 32B Instruct | 12 | AI2 |
| 151 | R1 1776 | 12 | Perplexity |
| 152 | Llama 3.2 90B (Vision) | 12 | Meta |
| 153 | EXAONE 4.0 32B | 12 | LG AI Research |
| 154 | Ministral 3 3B | 11 | Mistral |
| 155 | DeepHermes 3 - Mistral 24B | 11 | Nous Research |
| 156 | Jamba 1.7 Large | 11 | AI21 Labs |
| 157 | Granite 4.0 H Small | 11 | IBM |
| 158 | Qwen3 Omni 30B A3B | 11 | Alibaba |
| 159 | Qwen3.5 0.8B | 11 | Alibaba |
| 160 | LFM2 24B A2B | 10 | Liquid AI |
| 161 | Phi-4 | 10 | Microsoft |
| 162 | Gemma 3 27B | 10 | |
| 163 | Nova Micro | 10 | Amazon |
| 164 | NVIDIA Nemotron Nano 12B | 10 | NVIDIA |
| 165 | Phi-4 Multimodal | 10 | Microsoft |
| 166 | Qwen3.5 0.8B | 10 | Alibaba |
| 167 | Jamba Reasoning 3B | 10 | AI21 Labs |
| 168 | Reka Flash 3 | 10 | Reka AI |
| 169 | Olmo 3 7B Think | 9 | AI2 |
| 170 | Ling-mini-2.0 | 9 | InclusionAI |
| 171 | Gemma 3 12B | 9 | |
| 172 | Llama 3.2 11B (Vision) | 9 | Meta |
| 173 | Phi-4 Mini | 8 | Microsoft |
| 174 | Exaone 4.0 1.2B | 8 | LG AI Research |
| 175 | Olmo 3 7B | 8 | AI2 |
| 176 | Exaone 4.0 1.2B | 8 | LG AI Research |
| 177 | LFM2.5-1.2B-Thinking | 8 | Liquid AI |
| 178 | Jamba 1.7 Mini | 8 | AI21 Labs |
| 179 | LFM2.5-1.2B-Instruct | 8 | Liquid AI |
| 180 | LFM2 2.6B | 8 | Liquid AI |
| 181 | Granite 4.0 H 1B | 8 | IBM |
| 182 | Gemma 3 270M | 8 | |
| 183 | Apertus 70B Instruct | 8 | Swiss AI |
| 184 | Granite 4.0 Micro | 8 | IBM |
| 185 | DeepHermes 3 - Llama 8B | 8 | Nous Research |
| 186 | Granite 4.0 1B | 7 | IBM |
| 187 | Molmo2-8B | 7 | AI2 |
| 188 | LFM2 8B A1B | 7 | Liquid AI |
| 189 | Gemma 3n E4B | 6 | |
| 190 | Gemma 3 4B | 6 | |
| 191 | LFM2.5-VL-1.6B | 6 | Liquid AI |
| 192 | Granite 4.0 350M | 6 | IBM |
| 193 | Apertus 8B Instruct | 6 | Swiss AI |
| 194 | Gemma 3 1B | 6 | |
| 195 | Granite 4.0 H 350M | 5 | IBM |
| 196 | Gemma 3n E2B | 5 |
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。
Intelligence Index 综合10项严格的评测基准，全面衡量AI模型能力，避免单一维度的过拟合。