Name: 智能体 AI 排行榜 - 最佳 Agent 模型对比 (2026-08-04 历史榜单)
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 ⇕	模型Model ⇕
› 🥇	Claude Opus 5 (High) Anthropic · Proprietary	12.10% ±1.40%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 6 确认成功Confirmed Success ▲16.42%±2.75% 赞扬/抱怨Praise vs Complaint ▲21.37%±5.40% 可控性Steerability ▲9.08%±2.52% Bash 恢复Bash Recovery ▲12.53%±0.79% 工具幻觉Tool Hallucination ▼1.09%±0.16% 会话数Sessions 19,135
› 🥈	Claude Fable 5 (High) Anthropic · Proprietary	11.69% ±2.37%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 9 确认成功Confirmed Success ▲11.24%±4.40% 赞扬/抱怨Praise vs Complaint ▲23.14%±8.30% 可控性Steerability ▲10.21%±4.91% Bash 恢复Bash Recovery ▲12.71%±3.45% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 24,070
› 🥉	Claude Opus 5 (Max) Anthropic · Proprietary	11.50% ±1.66%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 6 确认成功Confirmed Success ▲16.39%±3.12% 赞扬/抱怨Praise vs Complaint ▲20.28%±6.32% 可控性Steerability ▲6.39%±3.11% Bash 恢复Bash Recovery ▲13.31%±0.90% 工具幻觉Tool Hallucination ▼1.13%±0.16% 会话数Sessions 14,959
› 4	Kimi K3 (Max) 月之暗面 · Kimi K3 license	10.18% ±1.10%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 9 确认成功Confirmed Success ▲15.49%±2.14% 赞扬/抱怨Praise vs Complaint ▲19.71%±4.00% 可控性Steerability ▲7.50%±2.04% Bash 恢复Bash Recovery ▲7.04%±0.91% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 24,617
› 5	GPT 5.6 Sol (xHigh) OpenAI · Proprietary	10.09% ±1.69%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 11 确认成功Confirmed Success ▲8.97%±3.44% 赞扬/抱怨Praise vs Complaint ▲21.32%±6.29% 可控性Steerability ▲9.28%±3.36% Bash 恢复Bash Recovery ▲9.73%±1.21% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 17,508
› 6	Claude Opus 4.8 (Thinking) Anthropic · Proprietary	9.21% ±1.70%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 12 确认成功Confirmed Success ▲8.25%±2.88% 赞扬/抱怨Praise vs Complaint ▲21.18%±5.38% 可控性Steerability ▲9.04%±2.96% Bash 恢复Bash Recovery ▲8.91%±2.48% 工具幻觉Tool Hallucination ▼1.35%±3.09% 会话数Sessions 34,811
› 7	GPT 5.5 (xHigh) OpenAI · Proprietary	8.47% ±0.99%

综合排名Overall Rank — 名次区间Rank Spread 3 ↔ 12 确认成功Confirmed Success ▲5.58%±2.05% 赞扬/抱怨Praise vs Complaint ▲13.36%±3.57% 可控性Steerability ▲8.38%±1.79% Bash 恢复Bash Recovery ▲13.84%±1.18% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 45,559
› 8	Claude Opus 4.7 (Thinking) Anthropic · Proprietary	8.17% ±1.30%

综合排名Overall Rank — 名次区间Rank Spread 3 ↔ 13 确认成功Confirmed Success ▲6.32%±2.73% 赞扬/抱怨Praise vs Complaint ▲11.26%±4.53% 可控性Steerability ▲9.28%±2.63% Bash 恢复Bash Recovery ▲12.96%±1.10% 工具幻觉Tool Hallucination ▼1.04%±0.19% 会话数Sessions 35,797
› 9	GPT 5.5 (High) OpenAI · Proprietary	7.54% ±0.93%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 14 确认成功Confirmed Success ▲3.75%±1.94% 赞扬/抱怨Praise vs Complaint ▲11.48%±3.33% 可控性Steerability ▲9.18%±1.65% Bash 恢复Bash Recovery ▲12.15%±1.00% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 70,733
› 10	Claude Sonnet 5 (High) Anthropic · Proprietary	7.50% ±2.11%

综合排名Overall Rank — 名次区间Rank Spread 3 ↔ 17 确认成功Confirmed Success ▲6.61%±4.19% 赞扬/抱怨Praise vs Complaint ▲14.00%±7.62% 可控性Steerability ▲5.67%±4.42% Bash 恢复Bash Recovery ▲10.21%±1.54% 工具幻觉Tool Hallucination ▼1.01%±0.17% 会话数Sessions 25,071
› 11	Claude Opus 4.7 Anthropic · Proprietary	7.26% ±1.35%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 15 确认成功Confirmed Success ▲5.37%±2.81% 赞扬/抱怨Praise vs Complaint ▲10.96%±4.47% 可控性Steerability ▲9.34%±2.63% Bash 恢复Bash Recovery ▲9.49%±2.05% 工具幻觉Tool Hallucination ▼1.12%±0.16% 会话数Sessions 36,357
› 12	GLM 5.2 (Max) 智谱 ZAI · MIT	6.66% ±0.92%

综合排名Overall Rank — 名次区间Rank Spread 6 ↔ 16 确认成功Confirmed Success ▲9.06%±1.85% 赞扬/抱怨Praise vs Complaint ▲12.63%±3.31% 可控性Steerability ▲5.64%±1.62% Bash 恢复Bash Recovery ▲4.83%±1.13% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 48,329
› 13	Claude Opus 4.6 Anthropic · Proprietary	6.14% ±1.33%

综合排名Overall Rank — 名次区间Rank Spread 8 ↔ 19 确认成功Confirmed Success ▲3.66%±2.80% 赞扬/抱怨Praise vs Complaint ▲7.91%±4.29% 可控性Steerability ▲7.39%±2.55% Bash 恢复Bash Recovery ▲10.57%±1.98% 工具幻觉Tool Hallucination ▼1.16%±0.16% 会话数Sessions 35,546
› 14	Grok 4.5 SpaceXAI · Proprietary	5.56% ±1.30%

综合排名Overall Rank — 名次区间Rank Spread 9 ↔ 20 确认成功Confirmed Success ▲4.22%±2.88% 赞扬/抱怨Praise vs Complaint ▲6.06%±4.68% 可控性Steerability ▲4.97%±2.43% Bash 恢复Bash Recovery ▲11.36%±1.10% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 26,702
› 15	GPT 5.5 OpenAI · Proprietary	5.44% ±0.87%

综合排名Overall Rank — 名次区间Rank Spread 10 ↔ 19 确认成功Confirmed Success ▲2.67%±1.88% 赞扬/抱怨Praise vs Complaint ▲5.98%±2.98% 可控性Steerability ▲5.94%±1.59% Bash 恢复Bash Recovery ▲11.43%±0.94% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 71,889
› 16	GPT 5.4 (High) OpenAI · Proprietary	4.74% ±0.89%

综合排名Overall Rank — 名次区间Rank Spread 12 ↔ 20 确认成功Confirmed Success ▲4.63%±1.92% 赞扬/抱怨Praise vs Complaint ▲2.82%±3.08% 可控性Steerability ▲5.78%±1.69% Bash 恢复Bash Recovery ▲9.29%±0.97% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 71,081
› 17	GPT 5.6 Luna (xHigh) OpenAI · Proprietary	4.08% ±1.76%

综合排名Overall Rank — 名次区间Rank Spread 11 ↔ 21 确认成功Confirmed Success ▲0.35%±3.94% 赞扬/抱怨Praise vs Complaint ▲6.00%±6.28% 可控性Steerability ▲2.23%±3.60% Bash 恢复Bash Recovery ▲10.65%±1.37% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 8,227
› 18	GPT 5.6 Terra (xHigh) OpenAI · Proprietary	3.51% ±1.35%

综合排名Overall Rank — 名次区间Rank Spread 13 ↔ 22 确认成功Confirmed Success ▲1.44%±3.22% 赞扬/抱怨Praise vs Complaint ▲3.87%±4.59% 可控性Steerability ▲4.64%±2.66% Bash 恢复Bash Recovery ▲9.33%±1.13% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 11,662
› 19	Claude Opus 4.8 Anthropic · Proprietary	3.09% ±2.11%

综合排名Overall Rank — 名次区间Rank Spread 13 ↔ 25 确认成功Confirmed Success ▲8.55%±2.90% 赞扬/抱怨Praise vs Complaint ▲13.54%±5.19% 可控性Steerability ▲7.97%±2.92% Bash 恢复Bash Recovery ▲9.25%±2.33% 工具幻觉Tool Hallucination ▼23.85%±7.21% 会话数Sessions 32,857
› 20	Claude Sonnet 4.6 Anthropic · Proprietary	2.99% ±1.30%

综合排名Overall Rank — 名次区间Rank Spread 15 ↔ 22 确认成功Confirmed Success ▲0.21%±2.86% 赞扬/抱怨Praise vs Complaint ▲0.72%±3.97% 可控性Steerability ▲2.62%±2.47% Bash 恢复Bash Recovery ▲10.36%±2.46% 工具幻觉Tool Hallucination ▼1.06%±0.20% 会话数Sessions 36,299
› 21	Deepseek V4 Flash (High) (20260731) DeepSeek · MIT	1.98% ±1.43%

综合排名Overall Rank — 名次区间Rank Spread 17 ↔ 26 确认成功Confirmed Success ▲6.99%±3.24% 赞扬/抱怨Praise vs Complaint ▲1.54%±4.79% 可控性Steerability ▲2.31%±3.30% Bash 恢复Bash Recovery ▲2.53%±1.22% 工具幻觉Tool Hallucination ▼1.15%±0.16% 会话数Sessions 12,528
› 22	Muse Spark 1.1 Meta · Proprietary	0.60% ±0.65%

综合排名Overall Rank — 名次区间Rank Spread 20 ↔ 29 确认成功Confirmed Success ▲5.88%±1.47% 赞扬/抱怨Praise vs Complaint ▲4.86%±1.91% 可控性Steerability ▲4.53%±1.23% Bash 恢复Bash Recovery ▲5.39%±1.40% 工具幻觉Tool Hallucination ▼1.13%±0.16% 会话数Sessions 57,855
› 23	Kimi K2.7 Code 月之暗面 · Modified MIT	0.30% ±1.90%

综合排名Overall Rank — 名次区间Rank Spread 18 ↔ 34 确认成功Confirmed Success ▲4.32%±3.95% 赞扬/抱怨Praise vs Complaint ▲2.09%±6.57% 可控性Steerability ▲3.26%±4.13% Bash 恢复Bash Recovery ▲2.83%±2.71% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 10,677
› 24	GLM 5.1 智谱 ZAI · MIT	0.30% ±0.77%

综合排名Overall Rank — 名次区间Rank Spread 20 ↔ 30 确认成功Confirmed Success ▲1.19%±1.72% 赞扬/抱怨Praise vs Complaint ▲0.19%±2.52% 可控性Steerability ▲0.83%±1.43% Bash 恢复Bash Recovery ▲0.59%±1.36% 工具幻觉Tool Hallucination ▼0.25%±0.29% 会话数Sessions 67,699
› 25	Qwen3.7 Max 阿里巴巴 · Proprietary	0.19% ±0.89%

综合排名Overall Rank — 名次区间Rank Spread 21 ↔ 31 确认成功Confirmed Success ▲0.43%±2.16% 赞扬/抱怨Praise vs Complaint ▲6.00%±2.77% 可控性Steerability ▲0.67%±1.60% Bash 恢复Bash Recovery ▲4.27%±1.46% 工具幻觉Tool Hallucination ▼0.53%±0.26% 会话数Sessions 25,831
› 26	Gemini 3.1 Pro Preview Google · Proprietary	0.45% ±0.71%

综合排名Overall Rank — 名次区间Rank Spread 22 ↔ 31 确认成功Confirmed Success ▲1.80%±1.56% 赞扬/抱怨Praise vs Complaint ▲2.67%±2.27% 可控性Steerability ▲3.10%±1.24% Bash 恢复Bash Recovery ▲10.87%±1.32% 工具幻觉Tool Hallucination ▼1.04%±0.20% 会话数Sessions 77,581
› 27	DeepSeek V4 Pro DeepSeek · MIT	0.47% ±0.88%

综合排名Overall Rank — 名次区间Rank Spread 22 ↔ 31 确认成功Confirmed Success ▲2.88%±2.20% 赞扬/抱怨Praise vs Complaint ▲4.26%±2.86% 可控性Steerability ▲0.61%±1.60% Bash 恢复Bash Recovery ▲3.83%±0.93% 工具幻觉Tool Hallucination ▼0.35%±0.28% 会话数Sessions 26,445
› 28	Gemini 3.5 Flash (High) Google · Proprietary	0.75% ±0.59%

综合排名Overall Rank — 名次区间Rank Spread 23 ↔ 31 确认成功Confirmed Success ▲0.25%±1.40% 赞扬/抱怨Praise vs Complaint ▲1.31%±1.83% 可控性Steerability ▲1.03%±1.07% Bash 恢复Bash Recovery ▲1.78%±0.97% 工具幻觉Tool Hallucination ▼0.12%±0.20% 会话数Sessions 86,749
› 29	Kimi K2.6 月之暗面 · Modified MIT	0.84% ±2.05%

综合排名Overall Rank — 名次区间Rank Spread 20 ↔ 36 确认成功Confirmed Success ▲2.67%±3.99% 赞扬/抱怨Praise vs Complaint ▲0.62%±6.51% 可控性Steerability ▲2.87%±4.30% Bash 恢复Bash Recovery ▲5.79%±3.75% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 10,783
› 30	Hy3 腾讯 · Apache 2.0	1.40% ±1.52%

综合排名Overall Rank — 名次区间Rank Spread 22 ↔ 36 确认成功Confirmed Success ▲2.71%±3.19% 赞扬/抱怨Praise vs Complaint ▲4.05%±5.24% 可控性Steerability ▲9.48%±2.78% Bash 恢复Bash Recovery ▲1.82%±2.01% 工具幻觉Tool Hallucination ▼0.67%±0.66% 会话数Sessions 13,416
› 31	Qwen3.7 Plus 阿里巴巴 · Proprietary	1.82% ±1.32%

综合排名Overall Rank — 名次区间Rank Spread 24 ↔ 36 确认成功Confirmed Success ▲0.53%±3.35% 赞扬/抱怨Praise vs Complaint ▲10.35%±4.01% 可控性Steerability ▲4.07%±2.78% Bash 恢复Bash Recovery ▲5.85%±1.78% 工具幻觉Tool Hallucination ▼0.01%±0.43% 会话数Sessions 15,388
› 32	Minimax M3 MiniMax · MiniMax Community License	2.31% ±0.86%

综合排名Overall Rank — 名次区间Rank Spread 28 ↔ 36 确认成功Confirmed Success ▲4.43%±2.23% 赞扬/抱怨Praise vs Complaint ▲9.54%±2.71% 可控性Steerability ▲3.54%±1.69% Bash 恢复Bash Recovery ▲5.31%±0.85% 工具幻觉Tool Hallucination ▼0.67%±0.31% 会话数Sessions 26,051
› 33	Mimo V2.5 Pro 小米 · MIT	2.44% ±0.94%

综合排名Overall Rank — 名次区间Rank Spread 28 ↔ 36 确认成功Confirmed Success ▲4.11%±2.27% 赞扬/抱怨Praise vs Complaint ▲7.18%±2.95% 可控性Steerability ▲2.35%±1.74% Bash 恢复Bash Recovery ▲1.47%±1.38% 工具幻觉Tool Hallucination ▼0.02%±0.35% 会话数Sessions 26,376
› 34	DeepSeek V4 Flash DeepSeek · MIT	2.81% ±0.89%

综合排名Overall Rank — 名次区间Rank Spread 29 ↔ 36 确认成功Confirmed Success ▲3.58%±2.37% 赞扬/抱怨Praise vs Complaint ▲9.92%±2.72% 可控性Steerability ▲2.23%±1.68% Bash 恢复Bash Recovery ▲2.85%±0.96% 工具幻觉Tool Hallucination ▼1.17%±0.39% 会话数Sessions 23,316
› 35	gemini-3.6-flash Google · Proprietary	3.01% ±1.53%

综合排名Overall Rank — 名次区间Rank Spread 28 ↔ 36 确认成功Confirmed Success ▲1.63%±3.66% 赞扬/抱怨Praise vs Complaint ▲5.19%±4.77% 可控性Steerability ▲6.70%±3.19% Bash 恢复Bash Recovery ▲2.69%±1.83% 工具幻觉Tool Hallucination ▼1.17%±0.16% 会话数Sessions 7,601
› 36	Gemini 3.5 Flash (Medium) Google · Proprietary	4.39% ±1.54%

综合排名Overall Rank — 名次区间Rank Spread 29 ↔ 37 确认成功Confirmed Success ▲10.48%±3.78% 赞扬/抱怨Praise vs Complaint ▲4.66%±4.69% 可控性Steerability ▲6.47%±2.98% Bash 恢复Bash Recovery ▲0.97%±2.46% 工具幻觉Tool Hallucination ▼0.60%±0.53% 会话数Sessions 11,436
› 37	Inkling Thinky · Apache 2.0	6.37% ±0.90%

综合排名Overall Rank — 名次区间Rank Spread 36 ↔ 38 确认成功Confirmed Success ▲9.72%±2.48% 赞扬/抱怨Praise vs Complaint ▲16.41%±2.63% 可控性Steerability ▲12.20%±1.80% Bash 恢复Bash Recovery ▲6.00%±1.11% 工具幻觉Tool Hallucination ▼0.50%±0.23% 会话数Sessions 30,904
› 38	Grok 4.3 (High) SpaceXAI · Proprietary	8.07% ±0.83%

综合排名Overall Rank — 名次区间Rank Spread 37 ↔ 41 确认成功Confirmed Success ▲8.97%±1.74% 赞扬/抱怨Praise vs Complaint ▲13.53%±1.98% 可控性Steerability ▲6.50%±1.27% Bash 恢复Bash Recovery ▲12.36%±2.66% 工具幻觉Tool Hallucination ▼0.99%±0.16% 会话数Sessions 58,240
› 39	Grok Build 0.1 SpaceXAI · Proprietary	8.39% ±0.87%

综合排名Overall Rank — 名次区间Rank Spread 38 ↔ 41 确认成功Confirmed Success ▲4.65%±1.80% 赞扬/抱怨Praise vs Complaint ▲10.71%±2.36% 可控性Steerability ▲10.14%±1.54% Bash 恢复Bash Recovery ▲17.28%±2.42% 工具幻觉Tool Hallucination ▼0.82%±0.16% 会话数Sessions 69,595
› 40	Gemini 3 Flash Google · Proprietary	8.83% ±0.77%

综合排名Overall Rank — 名次区间Rank Spread 38 ↔ 41 确认成功Confirmed Success ▲8.05%±1.62% 赞扬/抱怨Praise vs Complaint ▲11.11%±1.89% 可控性Steerability ▲4.71%±1.22% Bash 恢复Bash Recovery ▲20.16%±2.23% 工具幻觉Tool Hallucination ▼0.14%±0.86% 会话数Sessions 78,785
› 41	gemini-3.5-flash-lite Google · Proprietary	9.91% ±1.60%

综合排名Overall Rank — 名次区间Rank Spread 38 ↔ 42 确认成功Confirmed Success ▲14.53%±3.86% 赞扬/抱怨Praise vs Complaint ▲12.25%±4.37% 可控性Steerability ▲10.53%±3.16% Bash 恢复Bash Recovery ▲11.80%±3.22% 工具幻觉Tool Hallucination ▼0.45%±0.69% 会话数Sessions 7,530
› 42	Minimax M2.7 MiniMax · Modified MIT	11.55% ±1.12%

综合排名Overall Rank — 名次区间Rank Spread 41 ↔ 43 确认成功Confirmed Success ▲13.56%±2.56% 赞扬/抱怨Praise vs Complaint ▲16.13%±3.03% 可控性Steerability ▲13.76%±2.05% Bash 恢复Bash Recovery ▲15.26%±2.92% 工具幻觉Tool Hallucination ▼0.99%±0.20% 会话数Sessions 26,241
› 43	Nemotron 3 Ultra NVIDIA · OpenMDW-1.1	13.96% ±2.36%

综合排名Overall Rank — 名次区间Rank Spread 42 ↔ 45 确认成功Confirmed Success ▲16.07%±5.21% 赞扬/抱怨Praise vs Complaint ▲14.57%±6.75% 可控性Steerability ▲20.33%±4.80% Bash 恢复Bash Recovery ▲19.03%±5.81% 工具幻觉Tool Hallucination ▼0.22%±0.44% 会话数Sessions 11,411
› 44	Grok 4.3 SpaceXAI · Proprietary	14.44% ±1.09%

综合排名Overall Rank — 名次区间Rank Spread 43 ↔ 45 确认成功Confirmed Success ▲11.76%±1.67% 赞扬/抱怨Praise vs Complaint ▲16.24%±1.87% 可控性Steerability ▲6.62%±1.23% Bash 恢复Bash Recovery ▲38.63%±4.57% 工具幻觉Tool Hallucination ▼1.06%±0.16% 会话数Sessions 78,194
› 45	Gemma 4 31B Google · Apache 2.0	17.22% ±2.25%

综合排名Overall Rank — 名次区间Rank Spread 43 ↔ 45 确认成功Confirmed Success ▲0.72%±1.90% 赞扬/抱怨Praise vs Complaint ▲3.49%±2.80% 可控性Steerability ▲8.68%±1.74% Bash 恢复Bash Recovery ▲42.22%±7.84% 工具幻觉Tool Hallucination ▼30.99%±7.19% 会话数Sessions 56,152

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