数据来源：Artificial Analysis 模型榜单（aa_models_leaderboard.csv）· 原始 270 个模型，其中 4 个因缺少 AA 指数未纳入，实际展示 266 个有 AA 指数的模型，其中开源 176 个 / 闭源 90 个
指标口径：AA 指数为智能指数；任务平均成本为 $/task；输出速率 median tok/s；首字延迟与总响应时间单位为秒。缺失值以 — 表示。
默认按 AA 指数降序排列；点击表头可切换排序。
| 排名 ↕ | 模型名称 ↕ | 生产厂商 ↕ | 类型 ↕ | AA 指数 ↕ | 任务平均成本 ($/task) ↕ | 上下文长度 (tokens) ↕ | 输出速率 (tok/s) ↕ | 首字延迟 (s) ↕ | 总响应时间 (s) ↕ |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Claude Opus 5.5 (max with fallback) | Anthropic | 闭源 | 58 | 5.980 | 1M | 94.0 | 780.5 | 785.8 |
| 2 | Claude Opus 5.5 (xhigh with fallback) | Anthropic | 闭源 | 56 | 3.460 | 1M | 88.0 | 132.2 | 137.9 |
| 3 | Claude Opus 5.5 (high with fallback) | Anthropic | 闭源 | 54 | 1.820 | 1M | 81.0 | 29.9 | 36.0 |
| 4 | Claude Fable 5.1 (max with fallback) | Anthropic | 闭源 | 53 | 7.630 | 1M | 69.0 | 223.1 | 230.3 |
| 5 | Claude Fable 5.1 (xhigh with fallback) | Anthropic | 闭源 | 53 | 5.980 | 1M | 67.0 | 128.7 | 136.2 |
| 6 | GPT-6 Astra (max) | OpenAI | 闭源 | 53 | 3.260 | 1M | 63.0 | 292.2 | 300.1 |
| 7 | GPT-6 Astra (xhigh) | OpenAI | 闭源 | 52 | 2.310 | 1M | 57.0 | 101.3 | 110.1 |
| 8 | Claude Opus 5.5 (medium with fallback) | Anthropic | 闭源 | 51 | 1.340 | 1M | 82.0 | 14.3 | 20.4 |
| 9 | Claude Fable 5.1 (high with fallback) | Anthropic | 闭源 | 51 | 3.910 | 1M | 57.0 | 11.5 | 20.3 |
| 10 | GPT-6 Astra (high) | OpenAI | 闭源 | 51 | 1.730 | 1M | 60.0 | 38.0 | 46.4 |
| 11 | GPT-6 Astra (medium) | OpenAI | 闭源 | 50 | 1.540 | 1M | 57.0 | 4.8 | 13.5 |
| 12 | Claude Fable 5.1 (medium with fallback) | Anthropic | 闭源 | 49 | 2.980 | 1M | 58.0 | 7.3 | 15.9 |
| 13 | Muse Spark 1.3 (max) | Meta | 闭源 | 48 | 1.600 | 1M | 205.0 | 25.4 | 37.6 |
| 14 | GPT-6 Sol (max) | OpenAI | 闭源 | 48 | 1.060 | 872K | 87.0 | 112.7 | 118.5 |
| 15 | Claude Fable 5.1 (low with fallback) | Anthropic | 闭源 | 47 | 2.370 | 1M | 58.0 | 5.5 | 14.2 |
| 16 | Grok 4.7 (xhigh) | SpaceXAI | 闭源 | 46 | 3.740 | 500K | 83.0 | 54.4 | 60.5 |
| 17 | Grok 4.7 (high) | SpaceXAI | 闭源 | 46 | 2.730 | 500K | 83.0 | 44.1 | 50.1 |
| 18 | MiMo-V2.6-Pro | Xiaomi | 开源 | 46 | 0.130 | 1M | 44.0 | 3.8 | 60.0 |
| 19 | GPT-6 Astra (low) | OpenAI | 闭源 | 46 | 0.820 | 1M | 55.0 | 2.4 | 11.5 |
| 20 | Qwen3.8 Max (0902) | Alibaba | 闭源 | 45 | 5.410 | 984K | 39.0 | 3.0 | 66.6 |
| 21 | Muse Spark 1.3 (xhigh) | Meta | 闭源 | 45 | 1.370 | 1M | 261.0 | 25.2 | 34.8 |
| 22 | GLM-5.3 (max) | Z AI | 开源 | 45 | 2.010 | 1M | 87.0 | 2.6 | 31.4 |
| 23 | Grok 4.6 (high) | SpaceXAI | 闭源 | 44 | 1.860 | 500K | 66.0 | 28.5 | 36.0 |
| 24 | Grok 4.6 (xhigh) | SpaceXAI | 闭源 | 44 | 2.320 | 500K | 75.0 | 34.0 | 40.7 |
| 25 | GPT-6 Sol (xhigh) | OpenAI | 闭源 | 44 | 0.530 | 872K | 83.0 | 35.5 | 41.5 |
| 26 | Step 5 Preview | StepFun | 闭源 | 44 | 0.720 | 1M | 72.0 | 3.0 | 37.6 |
| 27 | Kimi K3 (max) | Kimi | 开源 | 44 | 2.000 | 1.05M | — | — | — |
| 28 | Grok 4.6 (medium) | SpaceXAI | 闭源 | 43 | 1.500 | 500K | 69.0 | 22.4 | 29.6 |
| 29 | GPT-6 Sol (high) | OpenAI | 闭源 | 43 | 0.370 | 872K | 83.0 | 9.1 | 15.2 |
| 30 | Claude Opus 5.5 (low with fallback) | Anthropic | 闭源 | 42 | 0.550 | 1M | 79.0 | 5.5 | 11.8 |
| 31 | GPT-5.6 Terra (max) | OpenAI | 闭源 | 42 | 1.400 | 1M | 102.0 | 133.4 | 138.3 |
| 32 | GLM-5.3-Flash | Z AI | 开源 | 42 | 0.250 | 1M | 55.0 | 3.1 | 48.5 |
| 33 | Gemini 3.8 Flash (high) | 闭源 | 41 | 1.240 | 1M | 285.0 | 29.8 | 31.5 | |
| 34 | Qwen3.8 2.4T A95B | Alibaba | 开源 | 40 | 2.160 | 984K | 39.0 | 3.1 | 67.8 |
| 35 | Qwen3.8-Flash-Next | Alibaba | 开源 | 40 | 0.370 | 256K | 57.0 | 2.5 | 46.4 |
| 36 | GPT-6 Sol (medium) | OpenAI | 闭源 | 40 | 0.250 | 872K | — | — | — |
| 37 | Gemini 3.8 Flash (medium) | 闭源 | 40 | 0.930 | 1M | — | — | — | |
| 38 | DeepSeek V4.1 Flash (max) | DeepSeek | 开源 | 39 | 0.270 | 1M | 221.0 | 1.1 | 12.4 |
| 39 | Claude Sonnet 5 (max) | Anthropic | 闭源 | 38 | 5.090 | 1M | 75.0 | 159.2 | 165.9 |
| 40 | GPT-5.6 Terra (xhigh) | OpenAI | 闭源 | 38 | 0.630 | 1M | 92.0 | 23.5 | 28.9 |
| 41 | MiMo-V2.6-Flash | Xiaomi | 开源 | 38 | 0.060 | 1M | 62.0 | 3.5 | 43.6 |
| 42 | GPT-6 Luna (max) | OpenAI | 闭源 | 37 | 0.070 | 1M | 163.0 | 90.6 | 93.7 |
| 43 | DeepSeek V4 Pro 0813 (max) | DeepSeek | 开源 | 36 | 0.670 | 1M | 100.0 | 1.7 | 26.6 |
| 44 | Grok 4.6 (low) | SpaceXAI | 闭源 | 35 | 0.480 | 500K | 64.0 | 6.3 | 14.1 |
| 45 | DeepSeek V4 Flash Vision (max) | DeepSeek | 闭源 | 35 | 0.310 | 1M | 223.0 | 1.1 | 12.4 |
| 46 | Kimi K3 (low) | Kimi | 开源 | 34 | — | 1.05M | — | — | — |
| 47 | Claude Sonnet 5 (xhigh) | Anthropic | 闭源 | 34 | 2.870 | 1M | 69.0 | 14.3 | 21.6 |
| 48 | GLM-5.3 (low) | Z AI | 开源 | 34 | 0.850 | 1M | 84.0 | 3.1 | 33.0 |
| 49 | GPT-5.6 Terra (high) | OpenAI | 闭源 | 34 | 0.340 | 1M | 91.0 | 2.8 | 8.2 |
| 50 | GPT-6 Sol (low) | OpenAI | 闭源 | 34 | 0.130 | 872K | 82.0 | 1.9 | 7.9 |
| 51 | GPT-6 Luna (xhigh) | OpenAI | 闭源 | 34 | 0.040 | 1M | 154.0 | 15.8 | 19.1 |
| 52 | Qwen3.8 27B (xhigh) | Alibaba | 开源 | 34 | 1.010 | 256K | 47.0 | 3.8 | 57.5 |
| 53 | Motif 3 | Motif Technologies | 开源 | 34 | — | 262K | — | — | — |
| 54 | Gemini 3.8 Flash (low) | 闭源 | 33 | — | 1M | — | — | — | |
| 55 | GPT-5.3 Codex (xhigh) | OpenAI | 闭源 | 33 | — | 400K | 118.0 | 67.4 | 71.7 |
| 56 | Motif 3 (Beta) | Motif Technologies | 闭源 | 32 | — | 262K | — | — | — |
| 57 | GPT-6 Luna (high) | OpenAI | 闭源 | 32 | 0.030 | 1M | 146.0 | 6.0 | 9.5 |
| 58 | Claude Sonnet 5 (high) | Anthropic | 闭源 | 32 | 1.790 | 1M | 65.0 | 3.9 | 11.6 |
| 59 | K2 Horizon 375B A23B | Institute of Foundation Models | 开源 | 31 | — | 524K | — | — | — |
| 60 | GPT-5.6 Terra (medium) | OpenAI | 闭源 | 30 | 0.180 | 1M | 85.0 | 1.6 | 7.5 |
| 61 | Gemini 3.1 Pro Preview | 闭源 | 30 | 0.670 | 1M | 118.0 | 25.0 | 29.3 | |
| 62 | GPT-6 Luna (medium) | OpenAI | 闭源 | 29 | 0.020 | 1M | — | — | — |
| 63 | MiniMax-M3 | MiniMax | 开源 | 29 | 0.510 | 1M | 174.0 | 1.1 | 15.4 |
| 64 | Nex-N2-Pro | Nex AGI | 开源 | 28 | — | 262K | — | — | — |
| 65 | Solar Pro 4 | Upstage | 闭源 | 28 | — | 512K | 93.0 | 1.9 | 28.8 |
| 66 | GPT-6 Sol (Non-reasoning) | OpenAI | 闭源 | 28 | 0.330 | 872K | 75.0 | 1.1 | 7.7 |
| 67 | Claude Sonnet 5 (medium) | Anthropic | 闭源 | 28 | 1.000 | 1M | 65.0 | 1.8 | 9.5 |
| 68 | Inkling Small | Thinking Machines | 开源 | 28 | — | 1M | 208.0 | 2.1 | 14.1 |
| 69 | Qwen3.8 27B (medium) | Alibaba | 开源 | 28 | 1.130 | 256K | 54.0 | 3.8 | 50.5 |
| 70 | GPT-5.6 Terra (low) | OpenAI | 闭源 | 27 | 0.140 | 1M | 91.0 | 1.6 | 7.1 |
| 71 | JT-4.1 Flash 236B A21B | China Mobile | 闭源 | 27 | — | 256K | — | — | — |
| 72 | Quasar 438B (max) | Multiverse Computing | 闭源 | 27 | 2.020 | 1M | 144.0 | 1.2 | 18.6 |
| 73 | Apodex 1.1 | Apodex | 闭源 | 26 | 0.460 | 256K | — | — | — |
| 74 | Qwen3.8 27B (low) | Alibaba | 开源 | 26 | 1.050 | 256K | 57.0 | 3.8 | 47.9 |
| 75 | GPT-5.5 Instant (June 2026) | OpenAI | 闭源 | 26 | 0.690 | 400K | 147.0 | 1.0 | 18.0 |
| 76 | MiMo-V2.5-Pro | Xiaomi | 开源 | 26 | 0.050 | 1M | 39.0 | 2.3 | 66.2 |
| 77 | Kimi K2.7 Code | Kimi | 开源 | 26 | 0.540 | 256K | — | — | — |
| 78 | K2 Horizon MoVA 36B A4B | Institute of Foundation Models | 开源 | 25 | — | 524K | — | — | — |
| 79 | Hy3 | Tencent | 开源 | 25 | 0.070 | 256K | 85.0 | 3.0 | 32.4 |
| 80 | MiMo-V2.5 | Xiaomi | 开源 | 25 | — | 1M | 44.0 | 25.9 | 82.6 |
| 81 | Qwen3.7 Plus | Alibaba | 闭源 | 25 | 0.220 | 1M | 75.0 | 2.2 | 35.7 |
| 82 | Inkling | Thinking Machines | 开源 | 25 | — | 1M | 194.0 | 1.8 | 14.7 |
| 83 | Ling 3.0 Flash | InclusionAI | 开源 | 25 | — | 262K | 373.0 | 3.0 | 9.7 |
| 84 | Solar Open2 250B | Upstage | 开源 | 25 | — | 1.05M | — | — | — |
| 85 | DeepSeek V4.1 Flash (Non-reasoning) | DeepSeek | 开源 | 25 | 0.150 | 1M | 239.0 | 1.1 | 3.2 |
| 86 | Ling-3.0-flash-VL | InclusionAI | 开源 | 25 | — | 262K | 145.0 | 1.9 | 19.1 |
| 87 | Claude Sonnet 5 (low) | Anthropic | 闭源 | 24 | 0.510 | 1M | 58.0 | 1.2 | 9.8 |
| 88 | Claude Sonnet 5 (Non-reasoning) | Anthropic | 闭源 | 23 | — | 1M | 66.0 | 1.1 | 8.7 |
| 89 | Nemotron 3 Ultra | NVIDIA | 开源 | 23 | 0.600 | 262K | 198.0 | 1.1 | 15.1 |
| 90 | A.X-K2 | SK Telecom | 开源 | 23 | — | 262K | — | — | — |
| 91 | Ling-3.0-flash-Fin | InclusionAI | 开源 | 23 | — | 262K | 161.0 | 1.7 | 17.3 |
| 92 | MiMo-V2-Flash (Feb 2026) | Xiaomi | 开源 | 22 | — | 256K | — | — | — |
| 93 | Gemini 3.5 Flash-Lite | 闭源 | 22 | 0.120 | 1M | 337.0 | 9.9 | 11.3 | |
| 94 | G9v3-39A5B | AI9Stars | 开源 | 22 | — | 131K | — | — | — |
| 95 | KAT-Coder-Pro V2 | KwaiKAT | 闭源 | 22 | — | 256K | — | — | — |
| 96 | Qwen3.5 397B A17B (Non-reasoning) | Alibaba | 开源 | 21 | — | 262K | 93.0 | 2.2 | 7.6 |
| 97 | GPT-6 Luna (low) | OpenAI | 闭源 | 21 | 0.004 | 1M | 147.0 | 2.0 | 5.4 |
| 98 | GPT-5.6 Terra (Non-reasoning) | OpenAI | 闭源 | 21 | 0.140 | 1M | 89.0 | 0.8 | 6.4 |
| 99 | K2 Horizon 7B | Institute of Foundation Models | 开源 | 21 | — | 524K | — | — | — |
| 100 | Qwen3.5 Omni Plus | Alibaba | 闭源 | 20 | — | 256K | 104.0 | 2.1 | 6.9 |
相同厂商颜色相同，横向条形图按指数排序。
双子图共用横轴（任务平均成本，对数轴）：上图为全部模型散点 + 红色实线全局帕累托前沿；下图为各厂商自身的帕累托前沿（虚线，非前沿点已剔除）。右上角下拉选择器同时控制两个子图，坐标轴范围固定不变。