Name: 智能体模型厂商排行榜（2026-09-13）- 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	13.90% ±1.99%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 4 确认成功Confirmed Success ▲23.70%±3.08% 赞扬/抱怨Praise vs Complaint ▲30.47%±7.84% 可控性Steerability ▲3.05%±4.64% Bash 恢复Bash Recovery ▲11.91%±0.82% 工具幻觉Tool Hallucination ▼0.37%±0.04% 会话数Sessions 12,800
› 🥈	OpenAIOpenAI GPT 6 Astra (Max) · Proprietary	11.90% ±2.77%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 6 确认成功Confirmed Success ▲19.54%±3.97% 赞扬/抱怨Praise vs Complaint ▲34.41%±10.56% 可控性Steerability ▲1.53%±6.56% Bash 恢复Bash Recovery ▲6.70%±1.19% 工具幻觉Tool Hallucination ▼0.38%±0.04% 会话数Sessions 9,958
› 🥉	月之暗面Moonshot Kimi K3 (Max) · Kimi K3 license	6.39% ±0.68%

综合排名Overall Rank — 名次区间Rank Spread 6 ↔ 12 确认成功Confirmed Success ▲14.80%±1.40% 赞扬/抱怨Praise vs Complaint ▲12.30%±2.39% 可控性Steerability ▲0.29%±1.37% Bash 恢复Bash Recovery ▲4.15%±0.55% 工具幻觉Tool Hallucination ▼0.38%±0.04% 会话数Sessions 104,470
› 4	腾讯Tencent Hy4 preview · Apache 2.0	4.96% ±1.13%

综合排名Overall Rank — 名次区间Rank Spread 7 ↔ 17 确认成功Confirmed Success ▲12.26%±2.18% 赞扬/抱怨Praise vs Complaint ▲8.52%±4.21% 可控性Steerability ▲2.92%±2.42% Bash 恢复Bash Recovery ▲6.72%±0.60% 工具幻觉Tool Hallucination ▼0.23%±0.07% 会话数Sessions 29,877
› 5	DeepSeekDeepSeek Deepseek V4.1 Flash (Max) · MIT	4.87% ±1.84%

综合排名Overall Rank — 名次区间Rank Spread 6 ↔ 20 确认成功Confirmed Success ▲13.75%±3.76% 赞扬/抱怨Praise vs Complaint ▲5.23%±6.30% 可控性Steerability ▲1.91%±4.61% Bash 恢复Bash Recovery ▲7.03%±1.00% 工具幻觉Tool Hallucination ▼0.26%±0.08% 会话数Sessions 8,444
› 6	智谱 ZAIZ.ai GLM 5.2 (Max) · MIT · SiliconFlow	4.52% ±0.74%

综合排名Overall Rank — 名次区间Rank Spread 9 ↔ 17 确认成功Confirmed Success ▲8.19%±1.59% 赞扬/抱怨Praise vs Complaint ▲10.41%±2.58% 可控性Steerability ▲3.04%±1.36% Bash 恢复Bash Recovery ▲0.57%±0.80% 工具幻觉Tool Hallucination ▼0.38%±0.04% 会话数Sessions 75,109
› 7	MetaMeta Muse Spark 1.3 (Max) · Proprietary	4.13% ±1.05%

综合排名Overall Rank — 名次区间Rank Spread 9 ↔ 20 确认成功Confirmed Success ▲11.56%±2.10% 赞扬/抱怨Praise vs Complaint ▲4.84%±3.59% 可控性Steerability ▲1.36%±2.36% Bash 恢复Bash Recovery ▲5.21%±0.94% 工具幻觉Tool Hallucination ▼0.38%±0.04% 会话数Sessions 23,122
› 8	阿里巴巴Alibaba Qwen3.8 Max · Proprietary	4.18% ±1.78%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 23 确认成功Confirmed Success ▲9.74%±3.45% 赞扬/抱怨Praise vs Complaint ▲8.44%±6.55% 可控性Steerability ▲0.57%±3.60% Bash 恢复Bash Recovery ▲2.95%±1.99% 工具幻觉Tool Hallucination ▼0.80%±0.38% 会话数Sessions 11,748
› 9	SpaceXAISpaceXAI Grok 4.5 · Proprietary	3.94% ±1.99%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 24 确认成功Confirmed Success ▲3.44%±4.23% 赞扬/抱怨Praise vs Complaint ▲0.94%±7.27% 可控性Steerability ▲5.37%±4.01% Bash 恢复Bash Recovery ▲9.63%±1.81% 工具幻觉Tool Hallucination ▼0.31%±0.06% 会话数Sessions 16,199
› 10	GoogleGoogle Gemini 3.6 Flash (High) · Proprietary	5.45% ±1.12%

综合排名Overall Rank — 名次区间Rank Spread 32 ↔ 39 确认成功Confirmed Success ▲5.96%±2.84% 赞扬/抱怨Praise vs Complaint ▲7.58%±3.32% 可控性Steerability ▲2.62%±2.27% Bash 恢复Bash Recovery ▲11.41%±1.69% 工具幻觉Tool Hallucination ▼0.32%±0.08% 会话数Sessions 20,191
› 11	小米Xiaomi Mimo V2.5 Pro · MIT	5.52% ±0.98%

综合排名Overall Rank — 名次区间Rank Spread 33 ↔ 39 确认成功Confirmed Success ▲7.84%±2.56% 赞扬/抱怨Praise vs Complaint ▲9.58%±2.90% 可控性Steerability ▲4.23%±1.89% Bash 恢复Bash Recovery ▲5.23%±1.72% 工具幻觉Tool Hallucination ▼0.71%±0.28% 会话数Sessions 39,850
› 12	MiniMaxMiniMax Minimax M3 · MiniMax Community License	5.63% ±0.96%

综合排名Overall Rank — 名次区间Rank Spread 33 ↔ 39 确认成功Confirmed Success ▲10.83%±2.59% 赞扬/抱怨Praise vs Complaint ▲11.92%±2.84% 可控性Steerability ▲7.14%±1.98% Bash 恢复Bash Recovery ▲1.70%±1.05% 工具幻觉Tool Hallucination ▼0.05%±0.12% 会话数Sessions 38,415
› 13	ThinkyThinky Inkling Small · Apache 2.0	7.84% ±1.55%

综合排名Overall Rank — 名次区间Rank Spread 34 ↔ 41 确认成功Confirmed Success ▲20.83%±4.53% 赞扬/抱怨Praise vs Complaint ▲18.42%±4.42% 可控性Steerability ▲7.45%±3.60% Bash 恢复Bash Recovery ▲7.84%±1.16% 工具幻觉Tool Hallucination ▼0.35%±0.37% 会话数Sessions 12,521
› 14	MistralMistral Mistral Medium 3.5 · Modified MIT	10.27% ±1.45%

综合排名Overall Rank — 名次区间Rank Spread 39 ↔ 41 确认成功Confirmed Success ▲16.74%±3.76% 赞扬/抱怨Praise vs Complaint ▲15.03%±3.96% 可控性Steerability ▲3.94%±3.22% Bash 恢复Bash Recovery ▲13.30%±2.44% 工具幻觉Tool Hallucination ▼2.34%±0.85% 会话数Sessions 9,986

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