Name: 智能体模型厂商排行榜（2026-09-16）- 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	13.71% ±1.72%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 2 确认成功Confirmed Success ▲19.83%±2.75% 赞扬/抱怨Praise vs Complaint ▲31.83%±6.86% 可控性Steerability ▲3.88%±3.61% Bash 恢复Bash Recovery ▲12.62%±0.76% 工具幻觉Tool Hallucination ▼0.37%±0.03% 会话数Sessions 13,320
› 🥈	OpenAIOpenAI GPT 6 Astra (Max) · Proprietary	11.54% ±2.10%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 6 确认成功Confirmed Success ▲17.70%±3.26% 赞扬/抱怨Praise vs Complaint ▲32.79%±8.37% 可控性Steerability ▲0.47%±4.67% Bash 恢复Bash Recovery ▲7.28%±1.11% 工具幻觉Tool Hallucination ▼0.37%±0.03% 会话数Sessions 10,372
› 🥉	月之暗面Moonshot Kimi K3 (Max) · Kimi K3 license	6.22% ±0.62%

综合排名Overall Rank — 名次区间Rank Spread 7 ↔ 13 确认成功Confirmed Success ▲11.91%±1.28% 赞扬/抱怨Praise vs Complaint ▲11.79%±2.18% 可控性Steerability ▲1.99%±1.28% Bash 恢复Bash Recovery ▲5.03%±0.52% 工具幻觉Tool Hallucination ▼0.37%±0.03% 会话数Sessions 108,615
› 4	腾讯Tencent Hy4 preview · Apache 2.0	5.01% ±1.02%

综合排名Overall Rank — 名次区间Rank Spread 7 ↔ 17 确认成功Confirmed Success ▲9.85%±1.98% 赞扬/抱怨Praise vs Complaint ▲8.76%±3.77% 可控性Steerability ▲1.23%±2.16% Bash 恢复Bash Recovery ▲7.44%±0.56% 工具幻觉Tool Hallucination ▼0.23%±0.06% 会话数Sessions 38,074
› 5	DeepSeekDeepSeek Deepseek V4.1 Flash (Max) · MIT	4.88% ±1.33%

综合排名Overall Rank — 名次区间Rank Spread 7 ↔ 19 确认成功Confirmed Success ▲13.35%±2.49% 赞扬/抱怨Praise vs Complaint ▲4.76%±4.56% 可控性Steerability ▲1.63%±3.40% Bash 恢复Bash Recovery ▲7.63%±0.71% 工具幻觉Tool Hallucination ▼0.29%±0.05% 会话数Sessions 20,080
› 6	GoogleGoogle Gemini 3.8 Flash (High) · Proprietary	8.27% ±3.71%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 13 确认成功Confirmed Success ▲13.14%±5.78% 赞扬/抱怨Praise vs Complaint ▲26.34%±13.87% 可控性Steerability ▲3.93%±9.60% Bash 恢复Bash Recovery ▲6.23%±2.11% 工具幻觉Tool Hallucination ▼0.41%±1.07% 会话数Sessions 3,618
› 7	智谱 ZAIZ.ai GLM 5.2 (Max) · MIT · SiliconFlow	4.37% ±0.69%

综合排名Overall Rank — 名次区间Rank Spread 9 ↔ 18 确认成功Confirmed Success ▲4.90%±1.50% 赞扬/抱怨Praise vs Complaint ▲9.76%±2.42% 可控性Steerability ▲4.88%±1.31% Bash 恢复Bash Recovery ▲1.93%±0.73% 工具幻觉Tool Hallucination ▼0.37%±0.03% 会话数Sessions 76,766
› 8	MetaMeta Muse Spark 1.3 (Max) · Proprietary	4.20% ±0.86%

综合排名Overall Rank — 名次区间Rank Spread 9 ↔ 20 确认成功Confirmed Success ▲10.31%±1.71% 赞扬/抱怨Praise vs Complaint ▲2.94%±2.94% 可控性Steerability ▲1.47%±1.94% Bash 恢复Bash Recovery ▲5.91%±0.77% 工具幻觉Tool Hallucination ▼0.37%±0.03% 会话数Sessions 31,052
› 9	阿里巴巴Alibaba Qwen3.8 Max · Proprietary	4.23% ±1.59%

综合排名Overall Rank — 名次区间Rank Spread 6 ↔ 23 确认成功Confirmed Success ▲6.43%±3.15% 赞扬/抱怨Praise vs Complaint ▲6.85%±5.83% 可控性Steerability ▲4.69%±3.20% Bash 恢复Bash Recovery ▲4.01%±1.88% 工具幻觉Tool Hallucination ▼0.83%±0.36% 会话数Sessions 13,213
› 10	SpaceXAISpaceXAI Grok 4.5 · Proprietary	2.92% ±0.99%

综合排名Overall Rank — 名次区间Rank Spread 12 ↔ 24 确认成功Confirmed Success ▲1.88%±2.27% 赞扬/抱怨Praise vs Complaint ▲0.10%±3.33% 可控性Steerability ▲5.86%±2.02% Bash 恢复Bash Recovery ▲6.59%±1.22% 工具幻觉Tool Hallucination ▼0.37%±0.03% 会话数Sessions 38,146
› 11	小米Xiaomi Mimo V2.5 Pro · MIT	5.73% ±0.87%

综合排名Overall Rank — 名次区间Rank Spread 35 ↔ 40 确认成功Confirmed Success ▲9.00%±2.20% 赞扬/抱怨Praise vs Complaint ▲9.96%±2.52% 可控性Steerability ▲3.99%±1.68% Bash 恢复Bash Recovery ▲4.93%±1.64% 工具幻觉Tool Hallucination ▼0.78%±0.25% 会话数Sessions 40,763
› 12	MiniMaxMiniMax Minimax M3 · MiniMax Community License	5.76% ±0.83%

综合排名Overall Rank — 名次区间Rank Spread 35 ↔ 40 确认成功Confirmed Success ▲12.04%±2.21% 赞扬/抱怨Praise vs Complaint ▲11.84%±2.44% 可控性Steerability ▲6.58%±1.68% Bash 恢复Bash Recovery ▲1.64%±0.90% 工具幻觉Tool Hallucination ▼0.01%±0.14% 会话数Sessions 39,156
› 13	ThinkyThinky Inkling Small · Apache 2.0	9.16% ±1.35%

综合排名Overall Rank — 名次区间Rank Spread 41 ↔ 43 确认成功Confirmed Success ▲20.37%±3.98% 赞扬/抱怨Praise vs Complaint ▲20.04%±3.68% 可控性Steerability ▲11.82%±3.16% Bash 恢复Bash Recovery ▲6.88%±1.20% 工具幻觉Tool Hallucination ▼0.46%±0.35% 会话数Sessions 12,659
› 14	MistralMistral Mistral Medium 3.5 · Modified MIT	10.88% ±1.46%

综合排名Overall Rank — 名次区间Rank Spread 41 ↔ 43 确认成功Confirmed Success ▲15.08%±3.71% 赞扬/抱怨Praise vs Complaint ▲15.27%±3.98% 可控性Steerability ▲6.01%±3.24% Bash 恢复Bash Recovery ▲15.80%±2.69% 工具幻觉Tool Hallucination ▼2.23%±0.75% 会话数Sessions 10,244
› 15	UpstageUpstage Solar Pro 4 · Proprietary	15.52% ±1.87%

综合排名Overall Rank — 名次区间Rank Spread 44 ↔ 46 确认成功Confirmed Success ▲20.07%±4.89% 赞扬/抱怨Praise vs Complaint ▲21.39%±4.65% 可控性Steerability ▲10.11%±3.83% Bash 恢复Bash Recovery ▲24.21%±4.40% 工具幻觉Tool Hallucination ▼1.83%±0.76% 会话数Sessions 8,897

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