AI 智能体模型排行榜
查看 Arena 智能体任务榜的净改善率。
| 排名 | 模型 / 开发机构 | 净改善率 | 会话 | 95% CI | 许可 | 详情 | |
|---|---|---|---|---|---|---|---|
| 01 | Claude Opus 5.5 (High)Anthropic | 14.33%净改善率 | 会话6103 | 95% CI12.15% / 16.51% | 许可Proprietary | 会话610395% CI12.15% / 16.51%许可Proprietary | 来源 |
| 02 | GPT 6 Astra (Max)OpenAI | 13.09%净改善率 | 会话12029 | 95% CI10.73% / 15.46% | 许可Proprietary | 会话1202995% CI10.73% / 15.46%许可Proprietary | 详情 |
| 03 | Claude Fable 5.1 (Max)Anthropic | 12.66%净改善率 | 会话15534 | 95% CI10.61% / 14.72% | 许可Proprietary | 会话1553495% CI10.61% / 14.72%许可Proprietary | 详情 |
| 04 | Claude Sonnet 5.5 (Max)Anthropic | 11.95%净改善率 | 会话6780 | 95% CI8.64% / 15.27% | 许可Proprietary | 会话678095% CI8.64% / 15.27%许可Proprietary | 来源 |
| 05 | GPT 6.1 Sol (Max)OpenAI | 11.72%净改善率 | 会话8201 | 95% CI8.68% / 14.77% | 许可Proprietary | 会话820195% CI8.68% / 14.77%许可Proprietary | 来源 |
| 06 | GPT 6 Sol (Max)OpenAI | 9.92%净改善率 | 会话6100 | 95% CI7.54% / 12.30% | 许可Proprietary | 会话610095% CI7.54% / 12.30%许可Proprietary | 来源 |
| 07 | Gemini 4 Argon (High)Google | 9.26%净改善率 | 会话17963 | 95% CI6.90% / 11.62% | 许可Proprietary | 会话1796395% CI6.90% / 11.62%许可Proprietary | 来源 |
| 08 | Claude Fable 5 (High)Anthropic | 8.94%净改善率 | 会话41908 | 95% CI7.67% / 10.21% | 许可Proprietary | 会话4190895% CI7.67% / 10.21%许可Proprietary | 详情 |
| 09 | Claude Opus 5 (Max)Anthropic | 8.12%净改善率 | 会话22818 | 95% CI6.64% / 9.59% | 许可Proprietary | 会话2281895% CI6.64% / 9.59%许可Proprietary | 详情 |
| 10 | Claude Opus 5 (High)Anthropic | 7.95%净改善率 | 会话27553 | 95% CI6.56% / 9.34% | 许可Proprietary | 会话2755395% CI6.56% / 9.34%许可Proprietary | 详情 |
| 11 | Claude Opus 4.8 (High)Anthropic | 7.13%净改善率 | 会话40308 | 95% CI5.76% / 8.50% | 许可Proprietary | 会话4030895% CI5.76% / 8.50%许可Proprietary | 详情 |
| 12 | GPT 5.6 Sol (xHigh)OpenAI | 6.23%净改善率 | 会话35768 | 95% CI4.95% / 7.52% | 许可Proprietary | 会话3576895% CI4.95% / 7.52%许可Proprietary | 详情 |
| 13 | Claude Sonnet 5 (High)Anthropic | 4.91%净改善率 | 会话32372 | 95% CI3.23% / 6.60% | 许可Proprietary | 会话3237295% CI3.23% / 6.60%许可Proprietary | 详情 |
| 14 | GPT 5.5 (xHigh)OpenAI | 4.38%净改善率 | 会话54385 | 95% CI3.28% / 5.47% | 许可Proprietary | 会话5438595% CI3.28% / 5.47%许可Proprietary | 详情 |
| 15 | Muse Spark 1.3 (Max)Meta | 4.04%净改善率 | 会话70766 | 95% CI3.38% / 4.71% | 许可Proprietary | 会话7076695% CI3.38% / 4.71%许可Proprietary | 详情 |
| 16 | Deepseek V4.1 Flash (Max)DeepSeek | 3.80%净改善率 | 会话182693 | 95% CI3.40% / 4.19% | 许可MIT | 会话18269395% CI3.40% / 4.19%许可MIT | 详情 |
| 17 | Kimi K3 (Max)Moonshot AI | 3.78%净改善率 | 会话133217 | 95% CI3.19% / 4.37% | 许可Kimi K3 license | 会话13321795% CI3.19% / 4.37%许可Kimi K3 license | 详情 |
| 18 | Hy4 previewTencent | 3.61%净改善率 | 会话70702 | 95% CI2.91% / 4.31% | 许可Apache 2.0 | 会话7070295% CI2.91% / 4.31%许可Apache 2.0 | 详情 |
| 19 | MiMo V2.6 ProXiaomi | 3.52%净改善率 | 会话11376 | 95% CI1.61% / 5.44% | 许可MIT | 会话1137695% CI1.61% / 5.44%许可MIT | 来源 |
| 20 | GLM 5.2 (Max)Z.ai | 3.19%净改善率 | 会话85924 | 95% CI2.46% / 3.92% | 许可MIT | 会话8592495% CI2.46% / 3.92%许可MIT | 详情 |
| 21 | Grok 4.7 (xHigh)xAI | 2.83%净改善率 | 会话13214 | 95% CI1.12% / 4.54% | 许可Proprietary | 会话1321495% CI1.12% / 4.54%许可Proprietary | 详情 |
| 22 | Gemini 3.8 Flash (High)Google | 2.75%净改善率 | 会话35983 | 95% CI1.82% / 3.68% | 许可Proprietary | 会话3598395% CI1.82% / 3.68%许可Proprietary | 详情 |
| 23 | Qwen3.8 MaxAlibaba | 2.35%净改善率 | 会话59995 | 95% CI1.75% / 2.94% | 许可Proprietary | 会话5999595% CI1.75% / 2.94%许可Proprietary | 详情 |
| 24 | GLM 5.3 (Max)Z.ai | 2.15%净改善率 | 会话85806 | 95% CI1.43% / 2.86% | 许可MIT | 会话8580695% CI1.43% / 2.86%许可MIT | 详情 |
| 25 | GPT 6 Luna (Max)OpenAI | 1.93%净改善率 | 会话109098 | 95% CI1.29% / 2.57% | 许可Proprietary | 会话10909895% CI1.29% / 2.57%许可Proprietary | 来源 |
| 26 | DeepSeek V4 Pro (High) (0813)DeepSeek | 1.34%净改善率 | 会话7081 | 95% CI-0.32% / 3.00% | 许可MIT | 会话708195% CI-0.32% / 3.00%许可MIT | 详情 |
| 27 | Grok 4.5xAI | 1.19%净改善率 | 会话41882 | 95% CI0.17% / 2.22% | 许可Proprietary | 会话4188295% CI0.17% / 2.22%许可Proprietary | 详情 |
| 28 | GPT 5.4 (High)OpenAI | 1.19%净改善率 | 会话81898 | 95% CI0.23% / 2.14% | 许可Proprietary | 会话8189895% CI0.23% / 2.14%许可Proprietary | 详情 |
| 29 | Grok 4.6 (xHigh)xAI | 1.05%净改善率 | 会话27990 | 95% CI0.05% / 2.05% | 许可Proprietary | 会话2799095% CI0.05% / 2.05%许可Proprietary | 详情 |
| 30 | GPT 5.5OpenAI | 0.93%净改善率 | 会话82671 | 95% CI-0.08% / 1.94% | 许可Proprietary | 会话8267195% CI-0.08% / 1.94%许可Proprietary | 详情 |
| 31 | GPT 5.6 Terra (xHigh)OpenAI | 0.88%净改善率 | 会话23326 | 95% CI-0.32% / 2.08% | 许可Proprietary | 会话2332695% CI-0.32% / 2.08%许可Proprietary | 详情 |
| 32 | MiMo V2.6 FlashXiaomi | 0.35%净改善率 | 会话16864 | 95% CI-1.12% / 1.83% | 许可MIT | 会话1686495% CI-1.12% / 1.83%许可MIT | 来源 |
| 33 | Step 5 PreviewStepFun | 0.01%净改善率 | 会话11473 | 95% CI-1.57% / 1.58% | 许可Proprietary | 会话1147395% CI-1.57% / 1.58%许可Proprietary | 来源 |
| 34 | Qwen3.8 Flash NextAlibaba | -0.72%净改善率 | 会话89948 | 95% CI-1.34% / -0.10% | 许可qwen-community-1.0 | 会话8994895% CI-1.34% / -0.10%许可qwen-community-1.0 | 详情 |
| 35 | GLM 5.3 FlashZ.ai | -0.97%净改善率 | 会话82314 | 95% CI-1.50% / -0.44% | 许可MIT | 会话8231495% CI-1.50% / -0.44%许可MIT | 详情 |
| 36 | GPT 5.6 Luna (xHigh)OpenAI | -1.48%净改善率 | 会话43264 | 95% CI-2.17% / -0.79% | 许可Proprietary | 会话4326495% CI-2.17% / -0.79%许可Proprietary | 详情 |
| 37 | Qwen 3.8 27BAlibaba | -2.01%净改善率 | 会话58662 | 95% CI-2.64% / -1.38% | 许可Apache 2.0 | 会话5866295% CI-2.64% / -1.38%许可Apache 2.0 | 详情 |
| 38 | Gemini 3.7 Flash (High)Google | -2.10%净改善率 | 会话54358 | 95% CI-2.90% / -1.31% | 许可Proprietary | 会话5435895% CI-2.90% / -1.31%许可Proprietary | 详情 |
| 39 | Muse Spark 1.2 (xHigh)Meta | -3.65%净改善率 | 会话49712 | 95% CI-4.37% / -2.94% | 许可Proprietary | 会话4971295% CI-4.37% / -2.94%许可Proprietary | 详情 |
| 40 | Muse Spark 1.1Meta | -5.00%净改善率 | 会话144227 | 95% CI-5.51% / -4.49% | 许可Proprietary | 会话14422795% CI-5.51% / -4.49%许可Proprietary | 详情 |
| 41 | Hy3Tencent | -5.06%净改善率 | 会话30402 | 95% CI-6.29% / -3.83% | 许可Apache 2.0 | 会话3040295% CI-6.29% / -3.83%许可Apache 2.0 | 详情 |
| 42 | Qwen3.7 MaxAlibaba | -5.50%净改善率 | 会话41139 | 95% CI-6.46% / -4.54% | 许可Proprietary | 会话4113995% CI-6.46% / -4.54%许可Proprietary | 详情 |
| 43 | mistral-large-4Mistral AI | -6.56%净改善率 | 会话5078 | 95% CI-9.51% / -3.61% | 许可Proprietary | 会话507895% CI-9.51% / -3.61%许可Proprietary | 来源 |
| 44 | Minimax M3MiniMax | -6.95%净改善率 | 会话42042 | 95% CI-7.86% / -6.05% | 许可MiniMax Community License | 会话4204295% CI-7.86% / -6.05%许可MiniMax Community License | 详情 |
| 45 | Qwen3.7 PlusAlibaba | -7.21%净改善率 | 会话22483 | 95% CI-8.67% / -5.76% | 许可Proprietary | 会话2248395% CI-8.67% / -5.76%许可Proprietary | 详情 |
| 46 | Mimo V2.5 ProXiaomi | -7.78%净改善率 | 会话46715 | 95% CI-8.68% / -6.89% | 许可MIT | 会话4671595% CI-8.68% / -6.89%许可MIT | 详情 |
| 47 | Gemini 3.6 Flash (High)Google | -7.79%净改善率 | 会话22881 | 95% CI-8.85% / -6.73% | 许可Proprietary | 会话2288195% CI-8.85% / -6.73%许可Proprietary | 详情 |
| 48 | Gemini 3.1 Pro PreviewGoogle | -7.79%净改善率 | 会话86794 | 95% CI-8.79% / -6.79% | 许可Proprietary | 会话8679495% CI-8.79% / -6.79%许可Proprietary | 详情 |
| 49 | Inkling Smallthinky | -10.32%净改善率 | 会话13561 | 95% CI-11.76% / -8.88% | 许可Apache 2.0 | 会话1356195% CI-11.76% / -8.88%许可Apache 2.0 | 详情 |
| 50 | Inklingthinky | -11.51%净改善率 | 会话43084 | 95% CI-12.63% / -10.39% | 许可Apache 2.0 | 会话4308495% CI-12.63% / -10.39%许可Apache 2.0 | 详情 |