Name: 智能体模型厂商排行榜（2026-07-21）- 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	12.72% ±2.00%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 4 确认成功Confirmed Success ▲10.67%±3.84% 赞扬/抱怨Praise vs Complaint ▲23.94%±7.42% 可控性Steerability ▲14.62%±3.80% Bash 恢复Bash Recovery ▲12.97%±1.30% 工具幻觉Tool Hallucination ▼1.39%±0.17% 会话数Sessions 23,549
› 🥈	OpenAIOpenAI GPT 5.6 Sol (xHigh) · Proprietary	10.12% ±1.69%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 8 确认成功Confirmed Success ▲7.25%±3.29% 赞扬/抱怨Praise vs Complaint ▲23.53%±6.57% 可控性Steerability ▲9.71%±2.78% Bash 恢复Bash Recovery ▲8.74%±1.30% 工具幻觉Tool Hallucination ▼1.39%±0.17% 会话数Sessions 15,991
› 🥉	月之暗面Moonshot Kimi K3 · Proprietary	9.71% ±1.52%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 9 确认成功Confirmed Success ▲14.00%±2.92% 赞扬/抱怨Praise vs Complaint ▲20.30%±5.45% 可控性Steerability ▲6.52%±3.14% Bash 恢复Bash Recovery ▲6.33%±1.27% 工具幻觉Tool Hallucination ▼1.39%±0.17% 会话数Sessions 11,490
› 4	智谱 ZAIZ.ai GLM 5.2 (Max) · MIT · SiliconFlow	6.50% ±1.00%

综合排名Overall Rank — 名次区间Rank Spread 6 ↔ 14 确认成功Confirmed Success ▲8.65%±1.97% 赞扬/抱怨Praise vs Complaint ▲12.94%±3.63% 可控性Steerability ▲4.71%±1.79% Bash 恢复Bash Recovery ▲4.78%±1.15% 工具幻觉Tool Hallucination ▼1.39%±0.17% 会话数Sessions 38,221
› 5	SpaceXAISpaceXAI Grok 4.5 · Proprietary	5.56% ±1.33%

综合排名Overall Rank — 名次区间Rank Spread 6 ↔ 15 确认成功Confirmed Success ▲3.86%±2.88% 赞扬/抱怨Praise vs Complaint ▲8.17%±4.91% 可控性Steerability ▲3.80%±2.34% Bash 恢复Bash Recovery ▲10.56%±1.14% 工具幻觉Tool Hallucination ▼1.39%±0.17% 会话数Sessions 21,424
› 6	MetaMeta Muse Spark 1.1 · Proprietary	0.67% ±0.89%

综合排名Overall Rank — 名次区间Rank Spread 17 ↔ 24 确认成功Confirmed Success ▲4.39%±2.03% 赞扬/抱怨Praise vs Complaint ▲4.30%±2.73% 可控性Steerability ▲4.50%±1.70% Bash 恢复Bash Recovery ▲6.40%±1.64% 工具幻觉Tool Hallucination ▼1.36%±0.17% 会话数Sessions 28,128
› 7	阿里巴巴Alibaba Qwen3.7 Max · Proprietary	0.09% ±1.07%

综合排名Overall Rank — 名次区间Rank Spread 17 ↔ 26 确认成功Confirmed Success ▲1.84%±2.60% 赞扬/抱怨Praise vs Complaint ▲5.73%±3.50% 可控性Steerability ▲0.02%±2.01% Bash 恢复Bash Recovery ▲7.20%±1.42% 工具幻觉Tool Hallucination ▼0.83%±0.29% 会话数Sessions 15,992
› 8	GoogleGoogle Gemini 3.1 Pro Preview · Proprietary	0.47% ±0.68%

综合排名Overall Rank — 名次区间Rank Spread 18 ↔ 26 确认成功Confirmed Success ▲2.05%±1.49% 赞扬/抱怨Praise vs Complaint ▲0.56%±2.22% 可控性Steerability ▲1.99%±1.22% Bash 恢复Bash Recovery ▲8.29%±1.11% 工具幻觉Tool Hallucination ▼1.32%±0.18% 会话数Sessions 67,658
› 9	DeepSeekDeepSeek DeepSeek V4 Pro · MIT	1.19% ±1.06%

综合排名Overall Rank — 名次区间Rank Spread 18 ↔ 27 确认成功Confirmed Success ▲4.80%±2.70% 赞扬/抱怨Praise vs Complaint ▲5.65%±3.42% 可控性Steerability ▲2.11%±2.05% Bash 恢复Bash Recovery ▲5.76%±1.02% 工具幻觉Tool Hallucination ▼0.87%±0.26% 会话数Sessions 16,514
› 10	腾讯Tencent Hy3 · Apache 2.0	2.23% ±2.87%

综合排名Overall Rank — 名次区间Rank Spread 18 ↔ 29 确认成功Confirmed Success ▲4.65%±6.11% 赞扬/抱怨Praise vs Complaint ▲2.87%±9.99% 可控性Steerability ▲7.10%±5.95% Bash 恢复Bash Recovery ▲2.56%±3.98% 工具幻觉Tool Hallucination ▼0.89%±0.82% 会话数Sessions 3,530
› 11	MiniMaxMiniMax Minimax M3 · MiniMax Community License	3.10% ±1.05%

综合排名Overall Rank — 名次区间Rank Spread 23 ↔ 29 确认成功Confirmed Success ▲7.49%±2.73% 赞扬/抱怨Praise vs Complaint ▲9.67%±3.33% 可控性Steerability ▲5.43%±2.14% Bash 恢复Bash Recovery ▲6.15%±0.95% 工具幻觉Tool Hallucination ▼0.93%±0.38% 会话数Sessions 16,030
› 12	小米Xiaomi Mimo V2.5 Pro · MIT	3.39% ±1.11%

综合排名Overall Rank — 名次区间Rank Spread 24 ↔ 29 确认成功Confirmed Success ▲5.88%±2.73% 赞扬/抱怨Praise vs Complaint ▲10.33%±3.35% 可控性Steerability ▲2.94%±2.14% Bash 恢复Bash Recovery ▲1.69%±1.91% 工具幻觉Tool Hallucination ▼0.49%±0.34% 会话数Sessions 16,479
› 13	ThinkyThinky Inkling · Apache 2.0	6.41% ±1.31%

综合排名Overall Rank — 名次区间Rank Spread 30 ↔ 33 确认成功Confirmed Success ▲7.19%±3.50% 赞扬/抱怨Praise vs Complaint ▲19.01%±3.70% 可控性Steerability ▲11.60%±3.00% Bash 恢复Bash Recovery ▲6.12%±1.58% 工具幻觉Tool Hallucination ▼0.40%±0.49% 会话数Sessions 10,678
› 14	NVIDIANvidia Nemotron 3 Ultra · OpenMDW-1.1	13.50% ±2.38%

综合排名Overall Rank — 名次区间Rank Spread 35 ↔ 38 确认成功Confirmed Success ▲15.08%±5.11% 赞扬/抱怨Praise vs Complaint ▲12.20%±7.28% 可控性Steerability ▲21.37%±4.80% Bash 恢复Bash Recovery ▲18.77%±5.33% 工具幻觉Tool Hallucination ▼0.09%±0.67% 会话数Sessions 10,263

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