Name: 智能体模型厂商排行榜（2026-09-30）- 最佳 Agent 模型对比 - 厂商聚合排名
Creator: LM Arena
License: https://creativecommons.org/licenses/by/4.0/
Keywords: AI 智能体模型排行榜,智能体 AI 排行榜,最佳 Agent 模型对比,AI 模型排行,智能体评测,净提升

排名Rank ⇕	厂商Lab ⇕
› 🥇	AnthropicAnthropic Claude Fable 5.1 (Max) · Proprietary	14.55% ±1.92%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 4 确认成功Confirmed Success ▲17.46%±2.90% 赞扬/抱怨Praise vs Complaint ▲33.73%±7.89% 可控性Steerability ▲8.95%±2.83% Bash 恢复Bash Recovery ▲12.31%±0.96% 工具幻觉Tool Hallucination ▼0.32%±0.03% 会话数Sessions 14,636
› 🥈	OpenAIOpenAI GPT 6 Astra (Max) · Proprietary	12.18% ±2.28%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 7 确认成功Confirmed Success ▲12.03%±4.03% 赞扬/抱怨Praise vs Complaint ▲34.70%±9.48% 可控性Steerability ▲8.24%±2.73% Bash 恢复Bash Recovery ▲5.59%±1.63% 工具幻觉Tool Hallucination ▼0.32%±0.03% 会话数Sessions 11,077
› 🥉	GoogleGoogle Gemini 4 Argon (High) · Proprietary	11.58% ±6.42%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 13 确认成功Confirmed Success ▲14.04%±9.96% 赞扬/抱怨Praise vs Complaint ▲38.08%±26.33% 可控性Steerability ▲12.36%±4.79% Bash 恢复Bash Recovery ▲6.85%±9.76% 工具幻觉Tool Hallucination ▼0.26%±0.04% 会话数Sessions 1,142
› 4	腾讯Tencent Hy4 preview · Apache 2.0	4.24% ±0.85%

综合排名Overall Rank — 名次区间Rank Spread 10 ↔ 20 确认成功Confirmed Success ▲6.37%±1.74% 赞扬/抱怨Praise vs Complaint ▲7.67%±3.20% 可控性Steerability ▲0.92%±1.07% Bash 恢复Bash Recovery ▲6.26%±0.55% 工具幻觉Tool Hallucination ▼0.00%±0.30% 会话数Sessions 53,732
› 5	MetaMeta Muse Spark 1.3 (Max) · Proprietary	4.19% ±0.59%

综合排名Overall Rank — 名次区间Rank Spread 10 ↔ 19 确认成功Confirmed Success ▲6.25%±1.29% 赞扬/抱怨Praise vs Complaint ▲3.27%±2.09% 可控性Steerability ▲5.50%±0.75% Bash 恢复Bash Recovery ▲5.62%±0.59% 工具幻觉Tool Hallucination ▼0.32%±0.03% 会话数Sessions 61,422
› 6	DeepSeekDeepSeek Deepseek V4.1 Flash (Max) · MIT	4.01% ±0.50%

综合排名Overall Rank — 名次区间Rank Spread 11 ↔ 19 确认成功Confirmed Success ▲8.65%±1.10% 赞扬/抱怨Praise vs Complaint ▲2.21%±1.83% 可控性Steerability ▲1.19%±0.69% Bash 恢复Bash Recovery ▲7.77%±0.33% 工具幻觉Tool Hallucination ▼0.20%±0.04% 会话数Sessions 95,889
› 7	月之暗面Moonshot Kimi K3 (Max) · Kimi K3 license	3.98% ±0.53%

综合排名Overall Rank — 名次区间Rank Spread 11 ↔ 20 确认成功Confirmed Success ▲7.32%±1.18% 赞扬/抱怨Praise vs Complaint ▲6.61%±1.89% 可控性Steerability ▲1.42%±0.73% Bash 恢复Bash Recovery ▲4.25%±0.50% 工具幻觉Tool Hallucination ▼0.32%±0.03% 会话数Sessions 129,908
› 8	SpaceXAISpaceXAI Grok 4.7 (xHigh) · Proprietary	4.13% ±3.18%

综合排名Overall Rank — 名次区间Rank Spread 4 ↔ 29 确认成功Confirmed Success ▲13.14%±5.76% 赞扬/抱怨Praise vs Complaint ▲3.57%±13.42% 可控性Steerability ▲3.68%±4.48% Bash 恢复Bash Recovery ▲7.37%±3.12% 工具幻觉Tool Hallucination ▼0.26%±0.04% 会话数Sessions 3,593
› 9	智谱 ZAIZ.ai GLM 5.2 (Max) · MIT · SiliconFlow	3.64% ±0.74%

综合排名Overall Rank — 名次区间Rank Spread 11 ↔ 23 确认成功Confirmed Success ▲5.23%±1.54% 赞扬/抱怨Praise vs Complaint ▲10.17%±2.67% 可控性Steerability ▲1.77%±0.95% Bash 恢复Bash Recovery ▲0.68%±0.83% 工具幻觉Tool Hallucination ▼0.32%±0.03% 会话数Sessions 83,747
› 10	阿里巴巴Alibaba Qwen3.8 Max · Proprietary	2.80% ±0.65%

综合排名Overall Rank — 名次区间Rank Spread 13 ↔ 24 确认成功Confirmed Success ▲5.24%±1.46% 赞扬/抱怨Praise vs Complaint ▲4.21%±2.35% 可控性Steerability ▲3.05%±0.86% Bash 恢复Bash Recovery ▲2.65%±0.71% 工具幻觉Tool Hallucination ▼1.13%±0.21% 会话数Sessions 49,157
› 11	MiniMaxMiniMax Minimax M3 · MiniMax Community License	6.71% ±0.89%

综合排名Overall Rank — 名次区间Rank Spread 36 ↔ 42 确认成功Confirmed Success ▲13.06%±2.49% 赞扬/抱怨Praise vs Complaint ▲14.31%±2.56% 可控性Steerability ▲6.93%±1.46% Bash 恢复Bash Recovery ▲0.87%±1.06% 工具幻觉Tool Hallucination ▼0.12%±0.18% 会话数Sessions 40,771
› 12	小米Xiaomi Mimo V2.5 Pro · MIT	7.23% ±0.91%

综合排名Overall Rank — 名次区间Rank Spread 37 ↔ 42 确认成功Confirmed Success ▲9.40%±2.34% 赞扬/抱怨Praise vs Complaint ▲11.77%±2.66% 可控性Steerability ▲7.43%±1.39% Bash 恢复Bash Recovery ▲6.54%±1.61% 工具幻觉Tool Hallucination ▼1.00%±0.27% 会话数Sessions 44,405
› 13	ThinkyThinky Inkling Small · Apache 2.0	9.80% ±1.42%

综合排名Overall Rank — 名次区间Rank Spread 41 ↔ 45 确认成功Confirmed Success ▲19.51%±4.33% 赞扬/抱怨Praise vs Complaint ▲21.63%±3.68% 可控性Steerability ▲13.12%±2.78% Bash 恢复Bash Recovery ▲5.64%±1.37% 工具幻觉Tool Hallucination ▼0.37%±0.29% 会话数Sessions 13,028
› 14	MistralMistral Mistral Medium 3.5 · Modified MIT	11.74% ±1.50%

综合排名Overall Rank — 名次区间Rank Spread 43 ↔ 45 确认成功Confirmed Success ▲13.52%±3.80% 赞扬/抱怨Praise vs Complaint ▲16.67%±3.93% 可控性Steerability ▲5.72%±2.23% Bash 恢复Bash Recovery ▲20.42%±3.34% 工具幻觉Tool Hallucination ▼2.35%±0.80% 会话数Sessions 10,797
› 15	UpstageUpstage Solar Pro 4 · Proprietary	15.77% ±1.91%

综合排名Overall Rank — 名次区间Rank Spread 46 ↔ 46 确认成功Confirmed Success ▲20.84%±5.48% 赞扬/抱怨Praise vs Complaint ▲22.11%±4.67% 可控性Steerability ▲9.10%±3.11% Bash 恢复Bash Recovery ▲25.09%±3.90% 工具幻觉Tool Hallucination ▼1.71%±0.66% 会话数Sessions 9,242

没有找到相关模型No matching models found

AI 智能体模型排行榜Agent Models Leaderboard 🏆 综合榜单