Name: 智能体模型厂商排行榜（2026-08-20）- 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 Opus 5 (High) · Proprietary	12.47% ±1.54%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 6 确认成功Confirmed Success ▲15.41%±3.16% 赞扬/抱怨Praise vs Complaint ▲20.52%±5.59% 可控性Steerability ▲11.15%±2.98% Bash 恢复Bash Recovery ▲14.23%±0.81% 工具幻觉Tool Hallucination ▼1.04%±0.17% 会话数Sessions 19,787
› 🥈	月之暗面Moonshot Kimi K3 (Max) · Kimi K3 license	10.41% ±0.62%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 6 确认成功Confirmed Success ▲17.97%±1.23% 赞扬/抱怨Praise vs Complaint ▲18.31%±2.23% 可控性Steerability ▲5.96%±1.13% Bash 恢复Bash Recovery ▲8.66%±0.54% 工具幻觉Tool Hallucination ▼1.14%±0.16% 会话数Sessions 83,274
› 🥉	OpenAIOpenAI GPT 5.6 Sol (xHigh) · Proprietary	9.74% ±1.39%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 11 确认成功Confirmed Success ▲10.00%±2.87% 赞扬/抱怨Praise vs Complaint ▲22.20%±5.11% 可控性Steerability ▲5.77%±2.88% Bash 恢复Bash Recovery ▲9.60%±1.03% 工具幻觉Tool Hallucination ▼1.14%±0.16% 会话数Sessions 26,310
› 4	DeepSeekDeepSeek DeepSeek V4 Pro (High) (0813) · MIT	6.26% ±1.28%

综合排名Overall Rank — 名次区间Rank Spread 8 ↔ 19 确认成功Confirmed Success ▲13.06%±2.65% 赞扬/抱怨Praise vs Complaint ▲3.82%±4.37% 可控性Steerability ▲2.35%±2.83% Bash 恢复Bash Recovery ▲10.94%±0.92% 工具幻觉Tool Hallucination ▼1.14%±0.16% 会话数Sessions 13,377
› 5	阿里巴巴Alibaba Qwen3.8 Max · Proprietary	6.20% ±1.36%

综合排名Overall Rank — 名次区间Rank Spread 8 ↔ 19 确认成功Confirmed Success ▲11.62%±3.03% 赞扬/抱怨Praise vs Complaint ▲6.81%±4.79% 可控性Steerability ▲4.00%±2.67% Bash 恢复Bash Recovery ▲8.39%±1.13% 工具幻觉Tool Hallucination ▼0.18%±0.33% 会话数Sessions 11,800
› 6	SpaceXAISpaceXAI Grok 4.5 · Proprietary	6.17% ±1.16%

综合排名Overall Rank — 名次区间Rank Spread 8 ↔ 19 确认成功Confirmed Success ▲5.47%±2.57% 赞扬/抱怨Praise vs Complaint ▲5.62%±4.11% 可控性Steerability ▲7.17%±2.19% Bash 恢复Bash Recovery ▲11.45%±1.14% 工具幻觉Tool Hallucination ▼1.14%±0.16% 会话数Sessions 30,936
› 7	智谱 ZAIZ.ai GLM 5.2 (Max) · MIT · SiliconFlow	5.82% ±0.87%

综合排名Overall Rank — 名次区间Rank Spread 10 ↔ 19 确认成功Confirmed Success ▲7.24%±1.87% 赞扬/抱怨Praise vs Complaint ▲9.42%±2.99% 可控性Steerability ▲5.22%±1.53% Bash 恢复Bash Recovery ▲6.09%±1.11% 工具幻觉Tool Hallucination ▼1.14%±0.16% 会话数Sessions 54,704
› 8	GoogleGoogle Gemini 3.7 Flash (High) · Proprietary	3.32% ±1.02%

综合排名Overall Rank — 名次区间Rank Spread 18 ↔ 25 确认成功Confirmed Success ▲9.96%±2.43% 赞扬/抱怨Praise vs Complaint ▲0.04%±3.35% 可控性Steerability ▲2.65%±2.05% Bash 恢复Bash Recovery ▲2.93%±1.14% 工具幻觉Tool Hallucination ▼1.10%±0.17% 会话数Sessions 18,109
› 9	MetaMeta Muse Spark 1.1 · Proprietary	0.93% ±0.61%

综合排名Overall Rank — 名次区间Rank Spread 24 ↔ 29 确认成功Confirmed Success ▲6.58%±1.43% 赞扬/抱怨Praise vs Complaint ▲5.57%±1.82% 可控性Steerability ▲3.39%±1.16% Bash 恢复Bash Recovery ▲5.92%±1.12% 工具幻觉Tool Hallucination ▼1.11%±0.16% 会话数Sessions 76,194
› 10	小米Xiaomi Mimo V2.5 Pro · MIT	2.25% ±0.88%

综合排名Overall Rank — 名次区间Rank Spread 29 ↔ 38 确认成功Confirmed Success ▲3.33%±2.20% 赞扬/抱怨Praise vs Complaint ▲7.79%±2.69% 可控性Steerability ▲1.73%±1.61% Bash 恢复Bash Recovery ▲1.54%±1.44% 工具幻觉Tool Hallucination ▼0.05%±0.31% 会话数Sessions 32,685
› 11	腾讯Tencent Hy3 · Apache 2.0	2.38% ±1.24%

综合排名Overall Rank — 名次区间Rank Spread 28 ↔ 38 确认成功Confirmed Success ▲1.87%±2.86% 赞扬/抱怨Praise vs Complaint ▲2.12%±4.15% 可控性Steerability ▲8.99%±2.37% Bash 恢复Bash Recovery ▲2.32%±1.99% 工具幻觉Tool Hallucination ▼1.25%±0.61% 会话数Sessions 19,741
› 12	MiniMaxMiniMax Minimax M3 · MiniMax Community License	3.15% ±0.83%

综合排名Overall Rank — 名次区间Rank Spread 32 ↔ 38 确认成功Confirmed Success ▲6.64%±2.23% 赞扬/抱怨Praise vs Complaint ▲9.34%±2.53% 可控性Steerability ▲6.16%±1.56% Bash 恢复Bash Recovery ▲5.76%±0.85% 工具幻觉Tool Hallucination ▼0.65%±0.29% 会话数Sessions 32,279
› 13	ThinkyThinky Inkling Small · Apache 2.0	6.82% ±1.75%

综合排名Overall Rank — 名次区间Rank Spread 38 ↔ 44 确认成功Confirmed Success ▲17.56%±4.91% 赞扬/抱怨Praise vs Complaint ▲14.84%±5.32% 可控性Steerability ▲12.81%±4.02% Bash 恢复Bash Recovery ▲10.73%±1.64% 工具幻觉Tool Hallucination ▼0.36%±0.48% 会话数Sessions 8,513
› 14	MistralMistral Mistral Medium 3.5 · Modified MIT	7.60% ±1.88%

综合排名Overall Rank — 名次区间Rank Spread 39 ↔ 45 确认成功Confirmed Success ▲11.54%±4.51% 赞扬/抱怨Praise vs Complaint ▲11.27%±5.36% 可控性Steerability ▲11.20%±3.86% Bash 恢复Bash Recovery ▲1.72%±3.07% 工具幻觉Tool Hallucination ▼2.29%±1.69% 会话数Sessions 6,018
› 15	UpstageUpstage Solar Pro 4 · Proprietary	10.82% ±2.39%

综合排名Overall Rank — 名次区间Rank Spread 40 ↔ 48 确认成功Confirmed Success ▲7.14%±5.80% 赞扬/抱怨Praise vs Complaint ▲17.18%±6.53% 可控性Steerability ▲16.03%±4.85% Bash 恢复Bash Recovery ▲12.78%±4.40% 工具幻觉Tool Hallucination ▼0.97%±2.93% 会话数Sessions 5,198
› 16	NVIDIANvidia Nemotron 3 Ultra · OpenMDW-1.1	15.36% ±2.34%

综合排名Overall Rank — 名次区间Rank Spread 47 ↔ 50 确认成功Confirmed Success ▲15.26%±4.99% 赞扬/抱怨Praise vs Complaint ▲15.46%±6.09% 可控性Steerability ▲20.47%±4.37% Bash 恢复Bash Recovery ▲25.88%±6.87% 工具幻觉Tool Hallucination ▼0.28%±0.41% 会话数Sessions 12,106

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