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 | Claude Opus 5.5Anthropic | 86.4BenchAlign | 推理82.4 | 编程83.5 | 多模态88.8 | 推理82.4编程83.5多模态88.8 | 详情 |
| 02 | GPT-6 AstraOpenAI | 84.8BenchAlign | 推理89.6 | 编程76.5 | 多模态— | 推理89.6编程76.5多模态— | 详情 |
| 03 | Claude Sonnet 5.5Anthropic | 83.9BenchAlign | 推理79.2 | 编程85.0 | 多模态— | 推理79.2编程85.0多模态— | 来源 |
| 04 | Gemini 4 ArgonGoogle | 83.4BenchAlign | 推理— | 编程72.8 | 多模态— | 推理—编程72.8多模态— | 来源 |
| 05 | Claude Fable 5.1Anthropic | 81.7BenchAlign | 推理81.4 | 编程77.8 | 多模态— | 推理81.4编程77.8多模态— | 详情 |
| 06 | GPT-6.1 SolOpenAI | 81.4BenchAlign | 推理86.8 | 编程71.0 | 多模态— | 推理86.8编程71.0多模态— | 来源 |
| 07 | Claude Opus 5Anthropic | 79.4BenchAlign | 推理77.3 | 编程71.7 | 多模态88.8 | 推理77.3编程71.7多模态88.8 | 详情 |
| 08 | Claude Fable 5Anthropic | 78.8BenchAlign | 推理— | 编程72.1 | 多模态— | 推理—编程72.1多模态— | 详情 |
| 09 | GPT-5.6 SolOpenAI | 77.9BenchAlign | 推理72.5 | 编程69.4 | 多模态88.6 | 推理72.5编程69.4多模态88.6 | 详情 |
| 10 | GPT-6 SolOpenAI | 75.4BenchAlign | 推理69.4 | 编程66.7 | 多模态— | 推理69.4编程66.7多模态— | 详情 |
| 11 | MiMo-V2.6-ProXiaomi | 74.1BenchAlign | 推理— | 编程63.8 | 多模态— | 推理—编程63.8多模态— | 来源 |
| 12 | Gemini 3.8 FlashGoogle | 73.2BenchAlign | 推理70.6 | 编程62.9 | 多模态83.9 | 推理70.6编程62.9多模态83.9 | 详情 |
| 13 | GPT-5.5 ProOpenAI | 72.5BenchAlign | 推理— | 编程— | 多模态— | 推理—编程—多模态— | 详情 |
| 14 | GPT-5.6 TerraOpenAI | 72.3BenchAlign | 推理65.5 | 编程64.2 | 多模态78.2 | 推理65.5编程64.2多模态78.2 | 详情 |
| 15 | Kimi K3Moonshot AI | 70.7BenchAlign | 推理65.8 | 编程60.6 | 多模态89.4 | 推理65.8编程60.6多模态89.4 | 详情 |
| 16 | Qwen3.8 MaxAlibaba | 70.4BenchAlign | 推理— | 编程54.5 | 多模态88.4 | 推理—编程54.5多模态88.4 | 详情 |
| 17 | Step 5 PreviewStepFun | 70.2BenchAlign | 推理— | 编程61.7 | 多模态61.2 | 推理—编程61.7多模态61.2 | 来源 |
| 18 | Claude Opus 4.8Anthropic | 69.1BenchAlign | 推理59.1 | 编程61.3 | 多模态87.7 | 推理59.1编程61.3多模态87.7 | 详情 |
| 19 | GPT-5.4 ProOpenAI | 68.9BenchAlign | 推理— | 编程— | 多模态— | 推理—编程—多模态— | 详情 |
| 20 | GLM-5.3Z.ai | 68.6BenchAlign | 推理77.1 | 编程57.9 | 多模态— | 推理77.1编程57.9多模态— | 详情 |
| 21 | Claude Opus 4.7 (Adaptive)Anthropic | 67.9BenchAlign | 推理— | 编程56.4 | 多模态50.1 | 推理—编程56.4多模态50.1 | 详情 |
| 22 | Grok 4.6xAI | 67.9BenchAlign | 推理57.7 | 编程60.5 | 多模态— | 推理57.7编程60.5多模态— | 详情 |
| 23 | GPT-5.5OpenAI | 67.7BenchAlign | 推理66.2 | 编程62.8 | 多模态71.4 | 推理66.2编程62.8多模态71.4 | 详情 |
| 24 | DeepSeek V4.1 FlashDeepSeek | 67.7BenchAlign | 推理— | 编程62.4 | 多模态— | 推理—编程62.4多模态— | 来源 |
| 25 | Gemini 3.7 FlashGoogle | 67.6BenchAlign | 推理— | 编程59.2 | 多模态83.6 | 推理—编程59.2多模态83.6 | 详情 |
| 26 | GPT-5.4OpenAI | 67.0BenchAlign | 推理60.6 | 编程48.7 | 多模态69.3 | 推理60.6编程48.7多模态69.3 | 详情 |
| 27 | Muse Spark 1.2Meta | 66.4BenchAlign | 推理— | 编程54.4 | 多模态— | 推理—编程54.4多模态— | 详情 |
| 28 | GPT-5.2 ProOpenAI | 66.2BenchAlign | 推理— | 编程— | 多模态— | 推理—编程—多模态— | 详情 |
| 29 | GPT-5.6 LunaOpenAI | 66.0BenchAlign | 推理54.5 | 编程63.4 | 多模态68.1 | 推理54.5编程63.4多模态68.1 | 详情 |
| 30 | Claude Sonnet 5Anthropic | 66.0BenchAlign | 推理— | 编程58.9 | 多模态78.4 | 推理—编程58.9多模态78.4 | 详情 |
| 31 | Muse Spark 1.1Meta | 65.9BenchAlign | 推理— | 编程54.1 | 多模态— | 推理—编程54.1多模态— | 详情 |
| 32 | GPT-6 LunaOpenAI | 65.6BenchAlign | 推理55.5 | 编程53.7 | 多模态— | 推理55.5编程53.7多模态— | 详情 |
| 33 | MiMo-V2.6-FlashXiaomi | 65.3BenchAlign | 推理— | 编程53.0 | 多模态— | 推理—编程53.0多模态— | 来源 |
| 34 | Gemini 3.1 ProGoogle | 64.6BenchAlign | 推理— | 编程41.0 | 多模态80.1 | 推理—编程41.0多模态80.1 | 详情 |
| 35 | Qwen3.8-Flash-NextAlibaba | 64.2BenchAlign | 推理— | 编程53.3 | 多模态84.1 | 推理—编程53.3多模态84.1 | 详情 |
| 36 | Claude Opus 4.7Anthropic | 64.1BenchAlign | 推理— | 编程56.5 | 多模态— | 推理—编程56.5多模态— | 详情 |
| 37 | Grok 4.5xAI | 63.9BenchAlign | 推理50.4 | 编程57.5 | 多模态— | 推理50.4编程57.5多模态— | 详情 |
| 38 | DeepSeek V4 Pro 0813DeepSeek | 63.9BenchAlign | 推理— | 编程48.9 | 多模态— | 推理—编程48.9多模态— | 详情 |
| 39 | Gemini 3.6 FlashGoogle | 63.2BenchAlign | 推理— | 编程52.6 | 多模态— | 推理—编程52.6多模态— | 详情 |
| 40 | dots3-note PreviewDots Studio | 63.1BenchAlign | 推理— | 编程— | 多模态— | 推理—编程—多模态— | 详情 |
| 41 | Ornith-1.5-397BOrnith AI | 62.5BenchAlign | 推理— | 编程55.1 | 多模态— | 推理—编程55.1多模态— | 详情 |
| 42 | Qwen3.7 MaxAlibaba | 62.5BenchAlign | 推理— | 编程45.3 | 多模态— | 推理—编程45.3多模态— | 详情 |
| 43 | Gemini 3.5 FlashGoogle | 62.4BenchAlign | 推理62.8 | 编程51.5 | 多模态87.8 | 推理62.8编程51.5多模态87.8 | 详情 |
| 44 | Claude Opus 4.6Anthropic | 62.0BenchAlign | 推理— | 编程48.0 | 多模态60.6 | 推理—编程48.0多模态60.6 | 详情 |
| 45 | GLM-5.2Z.ai | 61.5BenchAlign | 推理— | 编程55.2 | 多模态— | 推理—编程55.2多模态— | 详情 |
| 46 | Hy4 previewTencent | 60.9BenchAlign | 推理— | 编程58.0 | 多模态— | 推理—编程58.0多模态— | 详情 |
| 47 | GPT-5.3 CodexOpenAI | 60.8BenchAlign | 推理— | 编程55.6 | 多模态— | 推理—编程55.6多模态— | 详情 |
| 48 | GPT-5.2OpenAI | 60.3BenchAlign | 推理— | 编程38.2 | 多模态67.3 | 推理—编程38.2多模态67.3 | 详情 |
| 49 | Muse SparkMeta | 60.0BenchAlign | 推理— | 编程51.1 | 多模态78.5 | 推理—编程51.1多模态78.5 | 详情 |
| 50 | Gemini 3 ProGoogle | 59.4BenchAlign | 推理— | 编程43.6 | 多模态75.2 | 推理—编程43.6多模态75.2 | 详情 |