BenchLM 语言模型排行榜
按 BenchAlign 综合分比较语言模型,查看推理、编程与多模态表现。
| 排名 | 模型 / 开发机构 | BenchAlign | 推理 | 编程 | 多模态 | 详情 | |
|---|---|---|---|---|---|---|---|
| 01 | GPT-6 AstraOpenAI | 88.8BenchAlign | 推理89.6 | 编程74.0 | 多模态— | 推理89.6编程74.0多模态— | 详情 |
| 02 | Claude Opus 5.5Anthropic | 87.8BenchAlign | 推理82.4 | 编程83.3 | 多模态88.8 | 推理82.4编程83.3多模态88.8 | 详情 |
| 03 | Claude Sonnet 5.5Anthropic | 83.4BenchAlign | 推理79.2 | 编程85.6 | 多模态— | 推理79.2编程85.6多模态— | 来源 |
| 04 | Claude Fable 5.1Anthropic | 82.8BenchAlign | 推理81.4 | 编程79.7 | 多模态— | 推理81.4编程79.7多模态— | 详情 |
| 05 | Claude Opus 5Anthropic | 80.0BenchAlign | 推理77.3 | 编程72.0 | 多模态88.8 | 推理77.3编程72.0多模态88.8 | 详情 |
| 06 | Claude Fable 5Anthropic | 79.5BenchAlign | 推理— | 编程73.0 | 多模态— | 推理—编程73.0多模态— | 详情 |
| 07 | GPT-6 SolOpenAI | 79.2BenchAlign | 推理79.6 | 编程61.8 | 多模态— | 推理79.6编程61.8多模态— | 详情 |
| 08 | GPT-5.6 SolOpenAI | 79.0BenchAlign | 推理72.5 | 编程70.5 | 多模态88.6 | 推理72.5编程70.5多模态88.6 | 详情 |
| 09 | GPT-6.1 SolOpenAI | 77.6BenchAlign | 推理79.4 | 编程59.0 | 多模态— | 推理79.4编程59.0多模态— | 来源 |
| 10 | Gemini 4 ArgonGoogle | 77.2BenchAlign | 推理— | 编程74.0 | 多模态— | 推理—编程74.0多模态— | 来源 |
| 11 | MiMo-V2.6-ProXiaomi | 75.5BenchAlign | 推理— | 编程68.0 | 多模态— | 推理—编程68.0多模态— | 来源 |
| 12 | GPT-5.5 ProOpenAI | 74.6BenchAlign | 推理— | 编程— | 多模态— | 推理—编程—多模态— | 详情 |
| 13 | Gemini 3.8 FlashGoogle | 74.0BenchAlign | 推理70.6 | 编程63.6 | 多模态83.9 | 推理70.6编程63.6多模态83.9 | 详情 |
| 14 | GPT-5.6 TerraOpenAI | 73.2BenchAlign | 推理65.5 | 编程64.3 | 多模态78.2 | 推理65.5编程64.3多模态78.2 | 详情 |
| 15 | Kimi K3Moonshot AI | 72.1BenchAlign | 推理65.8 | 编程61.4 | 多模态89.4 | 推理65.8编程61.4多模态89.4 | 详情 |
| 16 | Qwen3.8 MaxAlibaba | 72.1BenchAlign | 推理— | 编程55.4 | 多模态88.4 | 推理—编程55.4多模态88.4 | 详情 |
| 17 | Claude Opus 4.8Anthropic | 70.8BenchAlign | 推理59.1 | 编程62.2 | 多模态87.7 | 推理59.1编程62.2多模态87.7 | 详情 |
| 18 | GPT-5.4 ProOpenAI | 70.7BenchAlign | 推理— | 编程— | 多模态— | 推理—编程—多模态— | 详情 |
| 19 | Step 5 PreviewStepFun | 70.4BenchAlign | 推理— | 编程61.4 | 多模态61.2 | 推理—编程61.4多模态61.2 | 来源 |
| 20 | Claude Opus 4.7 (Adaptive)Anthropic | 69.7BenchAlign | 推理— | 编程57.5 | 多模态50.1 | 推理—编程57.5多模态50.1 | 详情 |
| 21 | GPT-5.5OpenAI | 69.4BenchAlign | 推理66.2 | 编程62.7 | 多模态71.4 | 推理66.2编程62.7多模态71.4 | 详情 |
| 22 | Grok 4.6xAI | 69.4BenchAlign | 推理57.7 | 编程61.4 | 多模态— | 推理57.7编程61.4多模态— | 详情 |
| 23 | GPT-5.4OpenAI | 68.9BenchAlign | 推理60.6 | 编程49.6 | 多模态69.3 | 推理60.6编程49.6多模态69.3 | 详情 |
| 24 | GPT-5.2 ProOpenAI | 68.2BenchAlign | 推理— | 编程— | 多模态— | 推理—编程—多模态— | 详情 |
| 25 | Gemini 3.7 FlashGoogle | 68.0BenchAlign | 推理— | 编程59.3 | 多模态83.6 | 推理—编程59.3多模态83.6 | 详情 |
| 26 | Claude Sonnet 5Anthropic | 67.4BenchAlign | 推理— | 编程59.8 | 多模态78.4 | 推理—编程59.8多模态78.4 | 详情 |
| 27 | Muse Spark 1.1Meta | 66.6BenchAlign | 推理— | 编程55.3 | 多模态— | 推理—编程55.3多模态— | 详情 |
| 28 | MiMo-V2.6-FlashXiaomi | 66.4BenchAlign | 推理— | 编程53.1 | 多模态— | 推理—编程53.1多模态— | 来源 |
| 29 | GPT-5.6 LunaOpenAI | 66.2BenchAlign | 推理54.5 | 编程63.3 | 多模态68.1 | 推理54.5编程63.3多模态68.1 | 详情 |
| 30 | Claude Opus 4.7Anthropic | 65.8BenchAlign | 推理— | 编程57.9 | 多模态— | 推理—编程57.9多模态— | 详情 |
| 31 | Muse Spark 1.2Meta | 65.7BenchAlign | 推理— | 编程55.0 | 多模态— | 推理—编程55.0多模态— | 详情 |
| 32 | GLM-5.3Z.ai | 65.7BenchAlign | 推理77.1 | 编程56.6 | 多模态— | 推理77.1编程56.6多模态— | 详情 |
| 33 | Grok 4.5xAI | 65.7BenchAlign | 推理50.4 | 编程58.1 | 多模态— | 推理50.4编程58.1多模态— | 详情 |
| 34 | Gemini 3.6 FlashGoogle | 65.2BenchAlign | 推理— | 编程53.6 | 多模态— | 推理—编程53.6多模态— | 详情 |
| 35 | Gemini 3.1 ProGoogle | 65.1BenchAlign | 推理— | 编程42.3 | 多模态80.1 | 推理—编程42.3多模态80.1 | 详情 |
| 36 | GPT-6 LunaOpenAI | 65.0BenchAlign | 推理55.6 | 编程50.2 | 多模态— | 推理55.6编程50.2多模态— | 详情 |
| 37 | DeepSeek V4 Pro 0813DeepSeek | 65.0BenchAlign | 推理— | 编程49.2 | 多模态— | 推理—编程49.2多模态— | 详情 |
| 38 | DeepSeek V4.1 FlashDeepSeek | 64.7BenchAlign | 推理— | 编程58.8 | 多模态— | 推理—编程58.8多模态— | 来源 |
| 39 | Qwen3.8-Flash-NextAlibaba | 64.5BenchAlign | 推理— | 编程55.3 | 多模态84.1 | 推理—编程55.3多模态84.1 | 详情 |
| 40 | Claude Opus 4.6Anthropic | 64.3BenchAlign | 推理— | 编程49.5 | 多模态60.6 | 推理—编程49.5多模态60.6 | 详情 |
| 41 | Gemini 3.5 FlashGoogle | 64.0BenchAlign | 推理62.8 | 编程52.3 | 多模态87.8 | 推理62.8编程52.3多模态87.8 | 详情 |
| 42 | dots3-note PreviewDots Studio | 63.6BenchAlign | 推理— | 编程— | 多模态— | 推理—编程—多模态— | 详情 |
| 43 | Qwen3.7 MaxAlibaba | 63.5BenchAlign | 推理— | 编程45.4 | 多模态— | 推理—编程45.4多模态— | 详情 |
| 44 | GLM-5.2Z.ai | 63.2BenchAlign | 推理— | 编程56.2 | 多模态— | 推理—编程56.2多模态— | 详情 |
| 45 | Ornith-1.5-397BOrnith AI | 62.7BenchAlign | 推理— | 编程54.7 | 多模态— | 推理—编程54.7多模态— | 详情 |
| 46 | GPT-5.3 CodexOpenAI | 62.5BenchAlign | 推理— | 编程56.0 | 多模态— | 推理—编程56.0多模态— | 详情 |
| 47 | GPT-5.2OpenAI | 61.9BenchAlign | 推理— | 编程39.6 | 多模态67.3 | 推理—编程39.6多模态67.3 | 详情 |
| 48 | Gemini 3 ProGoogle | 61.6BenchAlign | 推理— | 编程45.4 | 多模态75.2 | 推理—编程45.4多模态75.2 | 详情 |
| 49 | Muse SparkMeta | 61.3BenchAlign | 推理— | 编程53.1 | 多模态78.5 | 推理—编程53.1多模态78.5 | 详情 |
| 50 | Hy4 previewTencent | 61.1BenchAlign | 推理— | 编程60.6 | 多模态— | 推理—编程60.6多模态— | 详情 |