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 Fable 5.1 (Max)Anthropic | 14.31%净改善率 | 会话15130 | 95% CI12.42% / 16.21% | 许可Proprietary | 会话1513095% CI12.42% / 16.21%许可Proprietary | 详情 |
| 02 | Claude Opus 5.5 (High)Anthropic | 13.82%净改善率 | 会话5359 | 95% CI11.65% / 15.99% | 许可Proprietary | 会话535995% CI11.65% / 15.99%许可Proprietary | 来源 |
| 03 | Claude Sonnet 5.5 (Max)Anthropic | 12.52%净改善率 | 会话5219 | 95% CI9.43% / 15.61% | 许可Proprietary | 会话521995% CI9.43% / 15.61%许可Proprietary | 来源 |
| 04 | GPT 6 Astra (Max)OpenAI | 12.27%净改善率 | 会话11717 | 95% CI10.04% / 14.50% | 许可Proprietary | 会话1171795% CI10.04% / 14.50%许可Proprietary | 详情 |
| 05 | GPT 6.1 Sol (Max)OpenAI | 11.23%净改善率 | 会话6278 | 95% CI8.46% / 13.99% | 许可Proprietary | 会话627895% CI8.46% / 13.99%许可Proprietary | 来源 |
| 06 | GPT 6 Sol (Max)OpenAI | 9.71%净改善率 | 会话6106 | 95% CI7.34% / 12.08% | 许可Proprietary | 会话610695% CI7.34% / 12.08%许可Proprietary | 来源 |
| 07 | Claude Opus 5 (High)Anthropic | 8.67%净改善率 | 会话27560 | 95% CI7.26% / 10.08% | 许可Proprietary | 会话2756095% CI7.26% / 10.08%许可Proprietary | 详情 |
| 08 | Claude Fable 5 (High)Anthropic | 8.21%净改善率 | 会话41832 | 95% CI7.00% / 9.43% | 许可Proprietary | 会话4183295% CI7.00% / 9.43%许可Proprietary | 详情 |
| 09 | Claude Opus 5 (Max)Anthropic | 7.92%净改善率 | 会话22781 | 95% CI6.45% / 9.40% | 许可Proprietary | 会话2278195% CI6.45% / 9.40%许可Proprietary | 详情 |
| 10 | Gemini 4 Argon (High)Google | 7.57%净改善率 | 会话10030 | 95% CI5.63% / 9.50% | 许可Proprietary | 会话1003095% CI5.63% / 9.50%许可Proprietary | 来源 |
| 11 | Claude Opus 4.8 (High)Anthropic | 6.64%净改善率 | 会话41132 | 95% CI5.30% / 7.98% | 许可Proprietary | 会话4113295% CI5.30% / 7.98%许可Proprietary | 详情 |
| 12 | GPT 5.6 Sol (xHigh)OpenAI | 6.45%净改善率 | 会话35896 | 95% CI5.17% / 7.73% | 许可Proprietary | 会话3589695% CI5.17% / 7.73%许可Proprietary | 详情 |
| 13 | Claude Sonnet 5 (High)Anthropic | 4.40%净改善率 | 会话32467 | 95% CI2.82% / 5.98% | 许可Proprietary | 会话3246795% CI2.82% / 5.98%许可Proprietary | 详情 |
| 14 | Kimi K3 (Max)Moonshot AI | 4.18%净改善率 | 会话132505 | 95% CI3.65% / 4.72% | 许可Kimi K3 license | 会话13250595% CI3.65% / 4.72%许可Kimi K3 license | 详情 |
| 15 | GPT 5.5 (xHigh)OpenAI | 4.15%净改善率 | 会话55006 | 95% CI3.13% / 5.18% | 许可Proprietary | 会话5500695% CI3.13% / 5.18%许可Proprietary | 详情 |
| 16 | Grok 4.7 (xHigh)xAI | 4.04%净改善率 | 会话11591 | 95% CI2.43% / 5.64% | 许可Proprietary | 会话1159195% CI2.43% / 5.64%许可Proprietary | 详情 |
| 17 | Deepseek V4.1 Flash (Max)DeepSeek | 4.02%净改善率 | 会话112938 | 95% CI3.56% / 4.49% | 许可MIT | 会话11293895% CI3.56% / 4.49%许可MIT | 详情 |
| 18 | Muse Spark 1.3 (Max)Meta | 4.00%净改善率 | 会话64671 | 95% CI3.40% / 4.59% | 许可Proprietary | 会话6467195% CI3.40% / 4.59%许可Proprietary | 详情 |
| 19 | Hy4 previewTencent | 3.95%净改善率 | 会话59388 | 95% CI3.16% / 4.75% | 许可Apache 2.0 | 会话5938895% CI3.16% / 4.75%许可Apache 2.0 | 详情 |
| 20 | GLM 5.2 (Max)Z.ai | 3.48%净改善率 | 会话86172 | 95% CI2.76% / 4.20% | 许可MIT | 会话8617295% CI2.76% / 4.20%许可MIT | 详情 |
| 21 | Gemini 3.8 Flash (High)Google | 2.97%净改善率 | 会话34396 | 95% CI2.12% / 3.81% | 许可Proprietary | 会话3439695% CI2.12% / 3.81%许可Proprietary | 详情 |
| 22 | Qwen3.8 MaxAlibaba | 2.51%净改善率 | 会话52824 | 95% CI1.89% / 3.14% | 许可Proprietary | 会话5282495% CI1.89% / 3.14%许可Proprietary | 详情 |
| 23 | GLM 5.3 (Max)Z.ai | 2.36%净改善率 | 会话86384 | 95% CI1.67% / 3.04% | 许可MIT | 会话8638495% CI1.67% / 3.04%许可MIT | 详情 |
| 24 | GPT 6 Luna (Max)OpenAI | 1.35%净改善率 | 会话29241 | 95% CI0.28% / 2.42% | 许可Proprietary | 会话2924195% CI0.28% / 2.42%许可Proprietary | 来源 |
| 25 | Grok 4.6 (xHigh)xAI | 1.28%净改善率 | 会话27380 | 95% CI0.28% / 2.28% | 许可Proprietary | 会话2738095% CI0.28% / 2.28%许可Proprietary | 详情 |
| 26 | Grok 4.5xAI | 1.21%净改善率 | 会话41920 | 95% CI0.20% / 2.22% | 许可Proprietary | 会话4192095% CI0.20% / 2.22%许可Proprietary | 详情 |
| 27 | DeepSeek V4 Pro (High) (0813)DeepSeek | 1.19%净改善率 | 会话7090 | 95% CI-0.47% / 2.84% | 许可MIT | 会话709095% CI-0.47% / 2.84%许可MIT | 详情 |
| 28 | GPT 5.4 (High)OpenAI | 1.05%净改善率 | 会话81597 | 95% CI0.13% / 1.98% | 许可Proprietary | 会话8159795% CI0.13% / 1.98%许可Proprietary | 详情 |
| 29 | GPT 5.5OpenAI | 1.04%净改善率 | 会话82416 | 95% CI0.06% / 2.02% | 许可Proprietary | 会话8241695% CI0.06% / 2.02%许可Proprietary | 详情 |
| 30 | Step 5 PreviewStepFun | 0.48%净改善率 | 会话9453 | 95% CI-0.99% / 1.95% | 许可Proprietary | 会话945395% CI-0.99% / 1.95%许可Proprietary | 来源 |
| 31 | GPT 5.6 Terra (xHigh)OpenAI | 0.40%净改善率 | 会话23457 | 95% CI-0.76% / 1.56% | 许可Proprietary | 会话2345795% CI-0.76% / 1.56%许可Proprietary | 详情 |
| 32 | GLM 5.3 FlashZ.ai | -0.44%净改善率 | 会话80836 | 95% CI-0.90% / 0.03% | 许可MIT | 会话8083695% CI-0.90% / 0.03%许可MIT | 详情 |
| 33 | MiMo V2.6 FlashXiaomi | -0.57%净改善率 | 会话15279 | 95% CI-1.85% / 0.72% | 许可MIT | 会话1527995% CI-1.85% / 0.72%许可MIT | 来源 |
| 34 | Qwen3.8 Flash NextAlibaba | -0.73%净改善率 | 会话83198 | 95% CI-1.36% / -0.10% | 许可qwen-community-1.0 | 会话8319895% CI-1.36% / -0.10%许可qwen-community-1.0 | 详情 |
| 35 | GPT 5.6 Luna (xHigh)OpenAI | -1.17%净改善率 | 会话42599 | 95% CI-1.84% / -0.49% | 许可Proprietary | 会话4259995% CI-1.84% / -0.49%许可Proprietary | 详情 |
| 36 | Gemini 3.7 Flash (High)Google | -1.49%净改善率 | 会话54594 | 95% CI-2.22% / -0.75% | 许可Proprietary | 会话5459495% CI-2.22% / -0.75%许可Proprietary | 详情 |
| 37 | Qwen 3.8 27BAlibaba | -1.66%净改善率 | 会话57343 | 95% CI-2.25% / -1.06% | 许可Apache 2.0 | 会话5734395% CI-2.25% / -1.06%许可Apache 2.0 | 详情 |
| 38 | Muse Spark 1.2 (xHigh)Meta | -3.27%净改善率 | 会话48150 | 95% CI-3.95% / -2.60% | 许可Proprietary | 会话4815095% CI-3.95% / -2.60%许可Proprietary | 详情 |
| 39 | Muse Spark 1.1Meta | -4.88%净改善率 | 会话142711 | 95% CI-5.33% / -4.44% | 许可Proprietary | 会话14271195% CI-5.33% / -4.44%许可Proprietary | 详情 |
| 40 | Qwen3.7 MaxAlibaba | -5.13%净改善率 | 会话40438 | 95% CI-6.06% / -4.20% | 许可Proprietary | 会话4043895% CI-6.06% / -4.20%许可Proprietary | 详情 |
| 41 | Hy3Tencent | -5.38%净改善率 | 会话29677 | 95% CI-6.59% / -4.17% | 许可Apache 2.0 | 会话2967795% CI-6.59% / -4.17%许可Apache 2.0 | 详情 |
| 42 | Minimax M3MiniMax | -6.87%净改善率 | 会话41267 | 95% CI-7.76% / -5.99% | 许可MiniMax Community License | 会话4126795% CI-7.76% / -5.99%许可MiniMax Community License | 详情 |
| 43 | Qwen3.7 PlusAlibaba | -7.15%净改善率 | 会话21722 | 95% CI-8.61% / -5.70% | 许可Proprietary | 会话2172295% CI-8.61% / -5.70%许可Proprietary | 详情 |
| 44 | Mimo V2.5 ProXiaomi | -7.52%净改善率 | 会话45239 | 95% CI-8.41% / -6.63% | 许可MIT | 会话4523995% CI-8.41% / -6.63%许可MIT | 详情 |
| 45 | Gemini 3.6 Flash (High)Google | -7.68%净改善率 | 会话22913 | 95% CI-8.74% / -6.63% | 许可Proprietary | 会话2291395% CI-8.74% / -6.63%许可Proprietary | 详情 |
| 46 | Gemini 3.1 Pro PreviewGoogle | -7.71%净改善率 | 会话86483 | 95% CI-8.69% / -6.73% | 许可Proprietary | 会话8648395% CI-8.69% / -6.73%许可Proprietary | 详情 |
| 47 | Inkling Smallthinky | -10.33%净改善率 | 会话13273 | 95% CI-11.73% / -8.92% | 许可Apache 2.0 | 会话1327395% CI-11.73% / -8.92%许可Apache 2.0 | 详情 |
| 48 | Inklingthinky | -10.86%净改善率 | 会话42813 | 95% CI-11.90% / -9.82% | 许可Apache 2.0 | 会话4281395% CI-11.90% / -9.82%许可Apache 2.0 | 详情 |
| 49 | Mistral Medium 3.5Mistral AI | -12.38%净改善率 | 会话11156 | 95% CI-13.90% / -10.86% | 许可Modified MIT | 会话1115695% CI-13.90% / -10.86%许可Modified MIT | 详情 |
| 50 | Solar Pro 4Upstage | -15.90%净改善率 | 会话9506 | 95% CI-17.75% / -14.05% | 许可Proprietary | 会话950695% CI-17.75% / -14.05%许可Proprietary | 详情 |