权威榜单
全球AI大模型能力排行，基于真实人类盲测投票与API调用数据
评测平台
2家
评测模块
6个
细分维度
26个
个人认为这个开源评测矩阵做得非常好，各项垂直领域的专业刷题跑分非常硬核，故特别添加了其开源模型的详细榜单作为矩阵补充。
🏆 综合总榜 数据看板
LLM Stats 综合大模型能力排名，基于 TrueSkill 算法
| 排名 | 模型名称 | 研发厂商 | LLM 综合总榜↑ | 开源总榜 |
|---|---|---|---|---|
| 1 |  | OpenAI | #1 | — |
| 2 |  | Anthropic | #2 | — |
| 3 | | Anthropic | #3 | — |
| 4 |  | Moonshot | #4 | 51 |
| 5 | | OpenAI | #5 | — |
| 6 |  | Meta | #6 | — |
| 7 | | Anthropic | #7 | — |
| 8 | | Anthropic | #8 | — |
| 9 |  | xAI | #9 | — |
| 10 | | OpenAI | #10 | — |
| 11 |  | ByteDance | #11 | — |
| 12 |  | Zhipu | #12 | 4 |
| 13 |  | Qwen | #13 | — |
| 14 | | Anthropic | #14 | — |
| 15 | | OpenAI | #15 | — |
| 16 | | Anthropic | #16 | — |
| 17 | | OpenAI | #17 | — |
| 18 | | Moonshot | #18 | 3 |
| 19 | | ByteDance | #19 | — |
| 20 |  | #20 | — | |
| 21 |  | DeepSeek | #21 | 6 |
| 22 | | #22 | — | |
| 23 | | Qwen | #23 | — |
| 24 |  | Tencent | #24 | 68 |
| 25 | | OpenAI | #25 | — |
| 26 |  | MiniMax | #26 | 17 |
| 27 | | OpenAI | #27 | — |
| 28 | | Meta | #28 | — |
| 29 | | OpenAI | #29 | — |
| 30 | | xAI | #30 | — |
| 31 | | Zhipu | #31 | 1 |
| 32 | | Qwen | #32 | — |
| 33 | | ByteDance | #33 | — |
| 34 | | DeepSeek | #34 | 12 |
| 35 | | Moonshot | #35 | 5 |
| 36 | | Qwen | #36 | 8 |
| 37 | | #37 | — | |
| 38 | | Anthropic | #38 | — |
| 39 | | Moonshot | #39 | 20 |
| 40 |  | Xiaomi | #40 | — |
| 41 | | Anthropic | #41 | — |
| 42 | | OpenAI | #42 | — |
| 43 | | Zhipu | #43 | 2 |
| 44 |  | StepFun | #44 | 30 |
| 45 | | #45 | — | |
| 46 | | MiniMax | #46 | 15 |
| 47 |  | Nvidia | #47 | 69 |
| 48 | | OpenAI | #48 | — |
| 49 | | OpenAI | #49 | — |
| 50 | | OpenAI | #50 | — |