Name: 智能体 AI 排行榜 - 最佳 Agent 模型对比 (2026-07-13 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 Fable 5 (High) Anthropic · Proprietary	13.94% ±1.56%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 2 确认成功Confirmed Success ▲17.27%±2.75% 赞扬/抱怨Praise vs Complaint ▲30.65%±5.67% 可控性Steerability ▲12.07%±2.94% Bash 恢复Bash Recovery ▲8.39%±1.81% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 16,059
› 🥈	GPT 5.6 Sol (xHigh) OpenAI · Proprietary	10.94% ±3.76%

综合排名Overall Rank — 名次区间Rank Spread 1 ↔ 9 确认成功Confirmed Success ▲10.93%±3.73% 赞扬/抱怨Praise vs Complaint ▲17.64%±7.48% 可控性Steerability ▲17.26%±16.35% Bash 恢复Bash Recovery ▲7.53%±1.68% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 7,881
› 🥉	Claude Opus 4.8 (Thinking) Anthropic · Proprietary	9.28% ±1.35%

综合排名Overall Rank — 名次区间Rank Spread 2 ↔ 7 确认成功Confirmed Success ▲7.82%±2.63% 赞扬/抱怨Praise vs Complaint ▲17.96%±4.91% 可控性Steerability ▲10.15%±2.48% Bash 恢复Bash Recovery ▲9.91%±1.00% 工具幻觉Tool Hallucination ▼0.56%±0.71% 会话数Sessions 33,392
› 4	GPT 5.5 (xHigh) OpenAI · Proprietary	8.26% ±0.87%

综合排名Overall Rank — 名次区间Rank Spread 2 ↔ 8 确认成功Confirmed Success ▲6.88%±1.71% 赞扬/抱怨Praise vs Complaint ▲12.61%±3.14% 可控性Steerability ▲6.70%±1.70% Bash 恢复Bash Recovery ▲13.79%±0.77% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 36,289
› 5	Claude Sonnet 5 (High) Anthropic · Proprietary	8.00% ±1.71%

综合排名Overall Rank — 名次区间Rank Spread 2 ↔ 13 确认成功Confirmed Success ▲10.33%±3.39% 赞扬/抱怨Praise vs Complaint ▲13.90%±6.50% 可控性Steerability ▲4.92%±3.24% Bash 恢复Bash Recovery ▲9.62%±0.88% 工具幻觉Tool Hallucination ▼1.21%±0.16% 会话数Sessions 23,640
› 6	Claude Opus 4.7 (Thinking) Anthropic · Proprietary	7.73% ±1.22%

综合排名Overall Rank — 名次区间Rank Spread 2 ↔ 13 确认成功Confirmed Success ▲5.22%±2.56% 赞扬/抱怨Praise vs Complaint ▲11.36%±4.35% 可控性Steerability ▲9.17%±2.31% Bash 恢复Bash Recovery ▲11.70%±1.16% 工具幻觉Tool Hallucination ▼1.22%±0.17% 会话数Sessions 34,357
› 7	Claude Opus 4.7 Anthropic · Proprietary	7.63% ±1.22%

综合排名Overall Rank — 名次区间Rank Spread 2 ↔ 13 确认成功Confirmed Success ▲5.28%±2.54% 赞扬/抱怨Praise vs Complaint ▲13.38%±4.39% 可控性Steerability ▲7.56%±2.27% Bash 恢复Bash Recovery ▲10.66%±1.32% 工具幻觉Tool Hallucination ▼1.27%±0.16% 会话数Sessions 34,950
› 8	GPT 5.5 (High) OpenAI · Proprietary	7.16% ±0.75%

综合排名Overall Rank — 名次区间Rank Spread 3 ↔ 13 确认成功Confirmed Success ▲6.42%±1.47% 赞扬/抱怨Praise vs Complaint ▲7.61%±2.61% 可控性Steerability ▲8.72%±1.42% Bash 恢复Bash Recovery ▲11.75%±1.00% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 61,455
› 9	GLM 5.2 (Max) 智谱 ZAI · MIT	6.24% ±1.10%

综合排名Overall Rank — 名次区间Rank Spread 4 ↔ 14 确认成功Confirmed Success ▲8.72%±2.15% 赞扬/抱怨Praise vs Complaint ▲12.59%±4.01% 可控性Steerability ▲4.10%±2.03% Bash 恢复Bash Recovery ▲4.45%±1.06% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 31,993
› 10	GPT 5.5 OpenAI · Proprietary	6.05% ±0.72%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 14 确认成功Confirmed Success ▲4.33%±1.46% 赞扬/抱怨Praise vs Complaint ▲5.53%±2.51% 可控性Steerability ▲7.99%±1.33% Bash 恢复Bash Recovery ▲11.06%±0.81% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 62,335
› 11	Claude Opus 4.6 Anthropic · Proprietary	5.88% ±1.23%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 14 确认成功Confirmed Success ▲1.67%±2.66% 赞扬/抱怨Praise vs Complaint ▲9.05%±4.17% 可控性Steerability ▲6.70%±2.25% Bash 恢复Bash Recovery ▲10.63%±1.33% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 34,075
› 12	GPT 5.4 (High) OpenAI · Proprietary	5.86% ±0.74%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 14 确认成功Confirmed Success ▲6.15%±1.49% 赞扬/抱怨Praise vs Complaint ▲4.40%±2.58% 可控性Steerability ▲7.88%±1.42% Bash 恢复Bash Recovery ▲9.55%±0.89% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 61,743
› 13	Grok 4.5 SpaceXAI · Proprietary	4.99% ±1.87%

综合排名Overall Rank — 名次区间Rank Spread 5 ↔ 14 确认成功Confirmed Success ▲6.95%±3.21% 赞扬/抱怨Praise vs Complaint ▲6.09%±5.58% 可控性Steerability ▲0.88%±6.12% Bash 恢复Bash Recovery ▲9.72%±1.34% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 9,851
› 14	Claude Opus 4.8 Anthropic · Proprietary	3.90% ±1.54%

综合排名Overall Rank — 名次区间Rank Spread 9 ↔ 15 确认成功Confirmed Success ▲6.75%±2.72% 赞扬/抱怨Praise vs Complaint ▲11.17%±4.77% 可控性Steerability ▲7.79%±2.58% Bash 恢复Bash Recovery ▲9.44%±1.22% 工具幻觉Tool Hallucination ▼15.63%±3.62% 会话数Sessions 31,463
› 15	Claude Sonnet 4.6 Anthropic · Proprietary	1.99% ±1.12%

综合排名Overall Rank — 名次区间Rank Spread 14 ↔ 17 确认成功Confirmed Success ▲1.95%±2.64% 赞扬/抱怨Praise vs Complaint ▲0.75%±3.63% 可控性Steerability ▲0.64%±2.13% Bash 恢复Bash Recovery ▲10.71%±1.39% 工具幻觉Tool Hallucination ▼1.30%±0.16% 会话数Sessions 34,873
› 16	GLM 5.1 智谱 ZAI · MIT	1.19% ±0.82%

综合排名Overall Rank — 名次区间Rank Spread 15 ↔ 19 确认成功Confirmed Success ▲0.78%±1.80% 赞扬/抱怨Praise vs Complaint ▲0.53%±2.83% 可控性Steerability ▲0.62%±1.68% Bash 恢复Bash Recovery ▲3.95%±0.86% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 51,362
› 17	Muse Spark 1.1 Meta · Proprietary	0.17% ±2.04%

综合排名Overall Rank — 名次区间Rank Spread 15 ↔ 26 确认成功Confirmed Success ▲2.06%±3.70% 赞扬/抱怨Praise vs Complaint ▲0.61%±5.26% 可控性Steerability ▲7.21%±6.54% Bash 恢复Bash Recovery ▲4.04%±3.44% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 9,262
› 18	Qwen3.7 Plus 阿里巴巴 · Proprietary	0.61% ±1.38%

综合排名Overall Rank — 名次区间Rank Spread 16 ↔ 25 确认成功Confirmed Success ▲1.10%±3.18% 赞扬/抱怨Praise vs Complaint ▲4.04%±4.31% 可控性Steerability ▲2.73%±3.04% Bash 恢复Bash Recovery ▲4.65%±1.73% 工具幻觉Tool Hallucination ▼0.19%±0.68% 会话数Sessions 10,048
› 19	Gemini 3.1 Pro Preview Google · Proprietary	0.70% ±0.69%

综合排名Overall Rank — 名次区间Rank Spread 17 ↔ 23 确认成功Confirmed Success ▲0.63%±1.53% 赞扬/抱怨Praise vs Complaint ▲0.07%±2.19% 可控性Steerability ▲1.85%±1.26% Bash 恢复Bash Recovery ▲7.21%±1.15% 工具幻觉Tool Hallucination ▼1.29%±0.15% 会话数Sessions 61,459
› 20	Kimi K2.7 Code 月之暗面 · Modified MIT	0.76% ±2.01%

综合排名Overall Rank — 名次区间Rank Spread 15 ↔ 25 确认成功Confirmed Success ▲4.51%±4.03% 赞扬/抱怨Praise vs Complaint ▲1.38%±7.14% 可控性Steerability ▲10.78%±4.26% Bash 恢复Bash Recovery ▲0.23%±2.20% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 7,468
› 21	Qwen3.7 Max 阿里巴巴 · Proprietary	0.81% ±1.39%

综合排名Overall Rank — 名次区间Rank Spread 16 ↔ 25 确认成功Confirmed Success ▲2.25%±3.30% 赞扬/抱怨Praise vs Complaint ▲5.38%±4.50% 可控性Steerability ▲4.16%±2.97% Bash 恢复Bash Recovery ▲6.84%±1.63% 工具幻觉Tool Hallucination ▼0.91%±0.28% 会话数Sessions 9,944
› 22	DeepSeek V4 Pro DeepSeek · MIT	1.11% ±1.36%

综合排名Overall Rank — 名次区间Rank Spread 17 ↔ 25 确认成功Confirmed Success ▲2.95%±3.30% 赞扬/抱怨Praise vs Complaint ▲4.79%±4.46% 可控性Steerability ▲3.10%±2.94% Bash 恢复Bash Recovery ▲4.50%±1.19% 工具幻觉Tool Hallucination ▼0.80%±0.34% 会话数Sessions 10,335
› 23	Gemini 3.5 Flash (High) Google · Proprietary	1.12% ±0.77%

综合排名Overall Rank — 名次区间Rank Spread 17 ↔ 25 确认成功Confirmed Success ▲1.66%±1.74% 赞扬/抱怨Praise vs Complaint ▲4.58%±2.42% 可控性Steerability ▲0.80%±1.43% Bash 恢复Bash Recovery ▲0.71%±1.29% 工具幻觉Tool Hallucination ▼1.17%±0.35% 会话数Sessions 45,991
› 24	Kimi K2.6 月之暗面 · Modified MIT	2.09% ±2.09%

综合排名Overall Rank — 名次区间Rank Spread 17 ↔ 26 确认成功Confirmed Success ▲2.36%±3.98% 赞扬/抱怨Praise vs Complaint ▲0.20%±6.64% 可控性Steerability ▲9.16%±4.32% Bash 恢复Bash Recovery ▲4.79%±4.00% 工具幻觉Tool Hallucination ▼1.33%±0.15% 会话数Sessions 7,620
› 25	Mimo V2.5 Pro 小米 · MIT	3.17% ±1.40%

综合排名Overall Rank — 名次区间Rank Spread 18 ↔ 26 确认成功Confirmed Success ▲6.00%±3.39% 赞扬/抱怨Praise vs Complaint ▲9.22%±4.41% 可控性Steerability ▲4.81%±3.16% Bash 恢复Bash Recovery ▲3.55%±1.76% 工具幻觉Tool Hallucination ▼0.63%±0.32% 会话数Sessions 10,379
› 26	Minimax M3 MiniMax · MiniMax Community License	3.23% ±1.36%

综合排名Overall Rank — 名次区间Rank Spread 18 ↔ 26 确认成功Confirmed Success ▲7.51%±3.38% 赞扬/抱怨Praise vs Complaint ▲8.13%±4.29% 可控性Steerability ▲7.92%±3.02% Bash 恢复Bash Recovery ▲6.27%±0.98% 工具幻觉Tool Hallucination ▼1.14%±0.20% 会话数Sessions 9,913
› 27	DeepSeek V4 Flash DeepSeek · MIT	4.64% ±1.30%

综合排名Overall Rank — 名次区间Rank Spread 23 ↔ 27 确认成功Confirmed Success ▲6.86%±3.44% 赞扬/抱怨Praise vs Complaint ▲12.90%±3.90% 可控性Steerability ▲6.99%±2.89% Bash 恢复Bash Recovery ▲4.10%±1.17% 工具幻觉Tool Hallucination ▼0.53%±0.48% 会话数Sessions 10,051
› 28	Gemini 3.5 Flash (Medium) Google · Proprietary	7.04% ±1.79%

综合排名Overall Rank — 名次区间Rank Spread 26 ↔ 30 确认成功Confirmed Success ▲11.84%±4.22% 赞扬/抱怨Praise vs Complaint ▲8.07%±5.27% 可控性Steerability ▲10.43%±3.28% Bash 恢复Bash Recovery ▲5.90%±4.10% 工具幻觉Tool Hallucination ▼1.04%±0.37% 会话数Sessions 7,041
› 29	Grok Build 0.1 SpaceXAI · Proprietary	8.21% ±0.79%

综合排名Overall Rank — 名次区间Rank Spread 27 ↔ 30 确认成功Confirmed Success ▲5.27%±1.79% 赞扬/抱怨Praise vs Complaint ▲13.65%±2.36% 可控性Steerability ▲12.84%±1.64% Bash 恢复Bash Recovery ▲9.83%±1.59% 工具幻觉Tool Hallucination ▼0.53%±0.17% 会话数Sessions 52,802
› 30	Grok 4.3 (High) SpaceXAI · Proprietary	8.31% ±0.80%

综合排名Overall Rank — 名次区间Rank Spread 27 ↔ 30 确认成功Confirmed Success ▲8.81%±1.76% 赞扬/抱怨Praise vs Complaint ▲15.63%±1.99% 可控性Steerability ▲7.97%±1.38% Bash 恢复Bash Recovery ▲10.06%±2.27% 工具幻觉Tool Hallucination ▼0.92%±0.17% 会话数Sessions 41,665
› 31	Gemini 3 Flash Google · Proprietary	8.54% ±0.77%

综合排名Overall Rank — 名次区间Rank Spread 27 ↔ 30 确认成功Confirmed Success ▲8.46%±1.58% 赞扬/抱怨Praise vs Complaint ▲12.83%±1.90% 可控性Steerability ▲5.24%±1.27% Bash 恢复Bash Recovery ▲16.46%±2.09% 工具幻觉Tool Hallucination ▼0.28%±1.26% 会话数Sessions 62,100
› 32	Minimax M2.7 MiniMax · Modified MIT	11.46% ±1.63%

综合排名Overall Rank — 名次区间Rank Spread 31 ↔ 33 确认成功Confirmed Success ▲16.97%±3.84% 赞扬/抱怨Praise vs Complaint ▲15.68%±4.65% 可控性Steerability ▲17.75%±3.43% Bash 恢复Bash Recovery ▲7.94%±3.22% 工具幻觉Tool Hallucination ▼1.06%±0.26% 会话数Sessions 10,183
› 33	Gemma 4 31B Google · Apache 2.0	13.58% ±1.42%

综合排名Overall Rank — 名次区间Rank Spread 31 ↔ 34 确认成功Confirmed Success ▲3.54%±1.68% 赞扬/抱怨Praise vs Complaint ▲6.03%±2.40% 可控性Steerability ▲5.33%±1.50% Bash 恢复Bash Recovery ▲30.16%±4.55% 工具幻觉Tool Hallucination ▼22.86%±4.41% 会话数Sessions 51,231
› 34	Nemotron 3 Ultra NVIDIA · OpenMDW-1.1	13.68% ±2.57%

综合排名Overall Rank — 名次区间Rank Spread 31 ↔ 34 确认成功Confirmed Success ▲13.69%±5.33% 赞扬/抱怨Praise vs Complaint ▲11.10%±7.85% 可控性Steerability ▲23.99%±5.09% Bash 恢复Bash Recovery ▲19.12%±5.92% 工具幻觉Tool Hallucination ▼0.51%±0.81% 会话数Sessions 9,511
› 35	Grok 4.3 SpaceXAI · Proprietary	15.31% ±1.01%

综合排名Overall Rank — 名次区间Rank Spread 32 ↔ 34 确认成功Confirmed Success ▲10.12%±1.62% 赞扬/抱怨Praise vs Complaint ▲16.64%±1.86% 可控性Steerability ▲8.96%±1.24% Bash 恢复Bash Recovery ▲41.90%±4.11% 工具幻觉Tool Hallucination ▼1.10%±0.17% 会话数Sessions 61,497

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