Name: 智能体模型厂商排行榜（2026-09-10）- 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	13.85% ±1.92%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 4 确认成功Confirmed Success ▲22.39%±3.04% 赞扬/抱怨Praise vs Complaint ▲36.35%±7.68% 可控性Steerability ▲0.91%±3.86% Bash 恢复Bash Recovery ▲11.07%±1.00% 工具幻觉Tool Hallucination ▼0.37%±0.04% 会话数Sessions 12,416
› 🥈	OpenAIOpenAI GPT 6 Astra (Max) · Proprietary	12.91% ±5.25%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 8 确认成功Confirmed Success ▲22.20%±7.09% 赞扬/抱怨Praise vs Complaint ▲45.41%±21.23% 可控性Steerability ▲10.10%±9.32% Bash 恢复Bash Recovery ▲6.73%±3.74% 工具幻觉Tool Hallucination ▼0.33%±0.06% 会话数Sessions 3,729
› 🥉	月之暗面Moonshot Kimi K3 (Max) · Kimi K3 license	6.46% ±0.70%

综合排名Overall Rank — 名次区间Rank Spread 6 ↔ 13 确认成功Confirmed Success ▲15.11%±1.43% 赞扬/抱怨Praise vs Complaint ▲12.49%±2.43% 可控性Steerability ▲0.02%±1.39% Bash 恢复Bash Recovery ▲4.29%±0.56% 工具幻觉Tool Hallucination ▼0.38%±0.04% 会话数Sessions 101,022
› 4	腾讯Tencent Hy4 preview · Apache 2.0	5.23% ±1.31%

综合排名Overall Rank — 名次区间Rank Spread 6 ↔ 18 确认成功Confirmed Success ▲11.82%±2.51% 赞扬/抱怨Praise vs Complaint ▲9.50%±4.85% 可控性Steerability ▲2.65%±2.83% Bash 恢复Bash Recovery ▲7.24%±0.67% 工具幻觉Tool Hallucination ▼0.25%±0.07% 会话数Sessions 19,613
› 5	智谱 ZAIZ.ai GLM 5.2 (Max) · MIT · SiliconFlow	4.56% ±0.74%

综合排名Overall Rank — 名次区间Rank Spread 9 ↔ 18 确认成功Confirmed Success ▲8.10%±1.58% 赞扬/抱怨Praise vs Complaint ▲10.71%±2.57% 可控性Steerability ▲3.16%±1.36% Bash 恢复Bash Recovery ▲0.46%±0.85% 工具幻觉Tool Hallucination ▼0.38%±0.04% 会话数Sessions 73,382
› 6	MetaMeta Muse Spark 1.3 (Max) · Proprietary	4.23% ±1.77%

综合排名Overall Rank — 名次区间Rank Spread 8 ↔ 23 确认成功Confirmed Success ▲8.37%±3.65% 赞扬/抱怨Praise vs Complaint ▲5.26%±5.80% 可控性Steerability ▲1.38%±4.47% Bash 恢复Bash Recovery ▲5.75%±1.31% 工具幻觉Tool Hallucination ▼0.38%±0.04% 会话数Sessions 8,770
› 7	DeepSeekDeepSeek DeepSeek V4 Pro (High) (0813) · MIT	4.14% ±0.92%

综合排名Overall Rank — 名次区间Rank Spread 9 ↔ 20 确认成功Confirmed Success ▲10.70%±1.87% 赞扬/抱怨Praise vs Complaint ▲6.03%±3.27% 可控性Steerability ▲1.61%±1.93% Bash 恢复Bash Recovery ▲5.20%±0.72% 工具幻觉Tool Hallucination ▼0.38%±0.04% 会话数Sessions 39,047
› 8	阿里巴巴Alibaba Qwen3.8 Max · Proprietary	4.66% ±1.98%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 22 确认成功Confirmed Success ▲10.84%±3.79% 赞扬/抱怨Praise vs Complaint ▲10.14%±7.34% 可控性Steerability ▲0.22%±3.91% Bash 恢复Bash Recovery ▲2.84%±2.32% 工具幻觉Tool Hallucination ▼0.72%±0.41% 会话数Sessions 10,071
› 9	GoogleGoogle Gemini 3.8 Flash (High) · Proprietary	7.19% ±3.77%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 20 确认成功Confirmed Success ▲11.70%±6.19% 赞扬/抱怨Praise vs Complaint ▲28.54%±13.11% 可控性Steerability ▲8.73%±11.31% Bash 恢复Bash Recovery ▲4.39%±2.56% 工具幻觉Tool Hallucination ▼0.08%±0.35% 会话数Sessions 2,863
› 10	SpaceXAISpaceXAI Grok 4.5 · Proprietary	4.29% ±2.03%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 25 确认成功Confirmed Success ▲4.03%±4.26% 赞扬/抱怨Praise vs Complaint ▲2.32%±7.52% 可控性Steerability ▲5.06%±4.02% Bash 恢复Bash Recovery ▲9.72%±1.80% 工具幻觉Tool Hallucination ▼0.33%±0.06% 会话数Sessions 15,863
› 11	MiniMaxMiniMax Minimax M3 · MiniMax Community License	5.38% ±0.95%

综合排名Overall Rank — 名次区间Rank Spread 33 ↔ 39 确认成功Confirmed Success ▲10.21%±2.57% 赞扬/抱怨Praise vs Complaint ▲11.54%±2.86% 可控性Steerability ▲6.82%±1.91% Bash 恢复Bash Recovery ▲1.66%±1.01% 工具幻觉Tool Hallucination ▼0.03%±0.13% 会话数Sessions 37,895
› 12	小米Xiaomi Mimo V2.5 Pro · MIT	5.42% ±0.98%

综合排名Overall Rank — 名次区间Rank Spread 33 ↔ 39 确认成功Confirmed Success ▲7.28%±2.54% 赞扬/抱怨Praise vs Complaint ▲10.16%±2.87% 可控性Steerability ▲3.78%±1.90% Bash 恢复Bash Recovery ▲5.17%±1.79% 工具幻觉Tool Hallucination ▼0.74%±0.30% 会话数Sessions 38,911
› 13	ThinkyThinky Inkling Small · Apache 2.0	7.09% ±1.58%

综合排名Overall Rank — 名次区间Rank Spread 33 ↔ 41 确认成功Confirmed Success ▲19.89%±4.54% 赞扬/抱怨Praise vs Complaint ▲17.32%±4.62% 可控性Steerability ▲5.97%±3.71% Bash 恢复Bash Recovery ▲8.11%±1.14% 工具幻觉Tool Hallucination ▼0.38%±0.38% 会话数Sessions 12,366
› 14	MistralMistral Mistral Medium 3.5 · Modified MIT	10.12% ±1.47%

综合排名Overall Rank — 名次区间Rank Spread 39 ↔ 41 确认成功Confirmed Success ▲16.44%±3.77% 赞扬/抱怨Praise vs Complaint ▲14.95%±4.01% 可控性Steerability ▲4.37%±3.28% Bash 恢复Bash Recovery ▲12.44%±2.43% 工具幻觉Tool Hallucination ▼2.41%±0.89% 会话数Sessions 9,879

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