Name: 智能体模型厂商排行榜（2026-08-12）- 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 (High) · Proprietary	12.04% ±2.54%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 9 确认成功Confirmed Success ▲10.77%±4.82% 赞扬/抱怨Praise vs Complaint ▲25.38%±8.93% 可控性Steerability ▲9.17%±5.13% Bash 恢复Bash Recovery ▲13.72%±3.63% 工具幻觉Tool Hallucination ▼1.18%±0.17% 会话数Sessions 24,390
› 🥈	OpenAIOpenAI GPT 5.6 Sol (xHigh) · Proprietary	10.72% ±1.79%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 10 确认成功Confirmed Success ▲9.82%±3.65% 赞扬/抱怨Praise vs Complaint ▲23.28%±6.64% 可控性Steerability ▲8.88%±3.65% Bash 恢复Bash Recovery ▲10.48%±1.24% 工具幻觉Tool Hallucination ▼1.18%±0.17% 会话数Sessions 18,050
› 🥉	月之暗面Moonshot Kimi K3 (Max) · Kimi K3 license	10.43% ±1.04%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 9 确认成功Confirmed Success ▲15.32%±2.05% 赞扬/抱怨Praise vs Complaint ▲20.17%±3.75% 可控性Steerability ▲7.54%±1.90% Bash 恢复Bash Recovery ▲7.92%±0.82% 工具幻觉Tool Hallucination ▼1.18%±0.17% 会话数Sessions 28,234
› 4	智谱 ZAIZ.ai GLM 5.2 (Max) · MIT · SiliconFlow	6.67% ±0.89%

综合排名Overall Rank — 名次区间Rank Spread 8 ↔ 17 确认成功Confirmed Success ▲8.43%±1.87% 赞扬/抱怨Praise vs Complaint ▲11.78%±3.15% 可控性Steerability ▲6.11%±1.57% Bash 恢复Bash Recovery ▲5.85%±1.10% 工具幻觉Tool Hallucination ▼1.18%±0.17% 会话数Sessions 52,849
› 5	SpaceXAISpaceXAI Grok 4.5 · Proprietary	5.73% ±1.20%

综合排名Overall Rank — 名次区间Rank Spread 8 ↔ 20 确认成功Confirmed Success ▲5.92%±2.70% 赞扬/抱怨Praise vs Complaint ▲4.37%±4.24% 可控性Steerability ▲6.43%±2.33% Bash 恢复Bash Recovery ▲10.78%±1.25% 工具幻觉Tool Hallucination ▼1.18%±0.17% 会话数Sessions 29,457
› 6	DeepSeekDeepSeek Deepseek V4 Flash (High) (20260731) · MIT	3.93% ±0.83%

综合排名Overall Rank — 名次区间Rank Spread 15 ↔ 22 确认成功Confirmed Success ▲9.14%±1.96% 赞扬/抱怨Praise vs Complaint ▲2.39%±2.74% 可控性Steerability ▲2.83%±1.63% Bash 恢复Bash Recovery ▲4.14%±0.71% 工具幻觉Tool Hallucination ▼1.17%±0.17% 会话数Sessions 33,134
› 7	MetaMeta Muse Spark 1.1 · Proprietary	1.09% ±0.60%

综合排名Overall Rank — 名次区间Rank Spread 20 ↔ 27 确认成功Confirmed Success ▲6.72%±1.40% 赞扬/抱怨Praise vs Complaint ▲4.99%±1.78% 可控性Steerability ▲3.16%±1.14% Bash 恢复Bash Recovery ▲5.74%±1.19% 工具幻觉Tool Hallucination ▼1.14%±0.17% 会话数Sessions 70,815
› 8	阿里巴巴Alibaba Qwen3.7 Max · Proprietary	0.01% ±0.85%

综合排名Overall Rank — 名次区间Rank Spread 21 ↔ 31 确认成功Confirmed Success ▲0.15%±2.10% 赞扬/抱怨Praise vs Complaint ▲4.90%±2.63% 可控性Steerability ▲0.20%±1.51% Bash 恢复Bash Recovery ▲4.67%±1.38% 工具幻觉Tool Hallucination ▼0.53%±0.26% 会话数Sessions 30,149
› 9	GoogleGoogle Gemini 3.5 Flash (High) · Proprietary	0.43% ±0.62%

综合排名Overall Rank — 名次区间Rank Spread 22 ↔ 31 确认成功Confirmed Success ▲0.49%±1.47% 赞扬/抱怨Praise vs Complaint ▲0.35%±1.93% 可控性Steerability ▲0.68%±1.11% Bash 恢复Bash Recovery ▲1.89%±0.99% 工具幻觉Tool Hallucination ▼0.28%±0.21% 会话数Sessions 93,367
› 10	腾讯Tencent Hy3 · Apache 2.0	1.30% ±1.28%

综合排名Overall Rank — 名次区间Rank Spread 23 ↔ 36 确认成功Confirmed Success ▲2.41%±2.82% 赞扬/抱怨Praise vs Complaint ▲2.27%±4.43% 可控性Steerability ▲8.52%±2.39% Bash 恢复Bash Recovery ▲3.49%±1.61% 工具幻觉Tool Hallucination ▼1.32%±0.72% 会话数Sessions 17,737
› 11	小米Xiaomi Mimo V2.5 Pro · MIT	2.21% ±0.88%

综合排名Overall Rank — 名次区间Rank Spread 29 ↔ 36 确认成功Confirmed Success ▲3.22%±2.19% 赞扬/抱怨Praise vs Complaint ▲7.10%±2.72% 可控性Steerability ▲2.37%±1.62% Bash 恢复Bash Recovery ▲1.70%±1.38% 工具幻觉Tool Hallucination ▼0.07%±0.34% 会话数Sessions 30,755
› 12	MiniMaxMiniMax Minimax M3 · MiniMax Community License	2.52% ±0.84%

综合排名Overall Rank — 名次区间Rank Spread 29 ↔ 36 确认成功Confirmed Success ▲5.76%±2.19% 赞扬/抱怨Praise vs Complaint ▲8.17%±2.60% 可控性Steerability ▲4.81%±1.59% Bash 恢复Bash Recovery ▲5.52%±0.83% 工具幻觉Tool Hallucination ▼0.64%±0.32% 会话数Sessions 30,528
› 13	ThinkyThinky Inkling · Apache 2.0	6.71% ±0.91%

综合排名Overall Rank — 名次区间Rank Spread 37 ↔ 39 确认成功Confirmed Success ▲12.50%±2.56% 赞扬/抱怨Praise vs Complaint ▲16.18%±2.61% 可控性Steerability ▲11.81%±1.82% Bash 恢复Bash Recovery ▲6.39%±1.06% 工具幻觉Tool Hallucination ▼0.52%±0.27% 会话数Sessions 35,265
› 14	MistralMistral Mistral Medium 3.5 · Modified MIT	6.93% ±2.00%

综合排名Overall Rank — 名次区间Rank Spread 36 ↔ 42 确认成功Confirmed Success ▲9.41%±4.64% 赞扬/抱怨Praise vs Complaint ▲9.83%±5.76% 可控性Steerability ▲11.23%±4.08% Bash 恢复Bash Recovery ▲1.23%±3.23% 工具幻觉Tool Hallucination ▼2.95%±2.22% 会话数Sessions 5,584
› 15	UpstageUpstage Solar Pro 4 · Proprietary	12.10% ±2.15%

综合排名Overall Rank — 名次区间Rank Spread 41 ↔ 46 确认成功Confirmed Success ▲7.28%±5.74% 赞扬/抱怨Praise vs Complaint ▲23.39%±5.20% 可控性Steerability ▲13.59%±4.08% Bash 恢复Bash Recovery ▲16.50%±5.38% 工具幻觉Tool Hallucination ▼0.26%±0.66% 会话数Sessions 4,150
› 16	NVIDIANvidia Nemotron 3 Ultra · OpenMDW-1.1	14.56% ±2.51%

综合排名Overall Rank — 名次区间Rank Spread 43 ↔ 47 确认成功Confirmed Success ▲15.84%±5.39% 赞扬/抱怨Praise vs Complaint ▲14.08%±6.73% 可控性Steerability ▲19.68%±4.72% Bash 恢复Bash Recovery ▲23.68%±7.02% 工具幻觉Tool Hallucination ▼0.47%±0.35% 会话数Sessions 11,935

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