Unverified50% confidenceBenchmarkExact time
即便是GPT-4级别的Agent，面对含有5步以上工具调用的复杂任务时，端到端成功率往往低于60%
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/25/2026
First Seen
Valid until: 10/23/2026
Sources
Related Claims
Unverified当链路节点超过5个时，Agent系统整体成功率往往不足60%77% similarUnverifiedAgents suffer from error accumulation in multi-step tasks; a ten-step sequential process where each step has 95% accuracy results in an overall success rate that drops to approximately 60%.74% similarUnverified斯坦福大学的AgentBench基准测试（2023年）量化证明，在复杂多步骤任务中纯ReAct范式的成功率往往不超过30%70% similarUnverifiedAfter Hermes Agent's Skill mechanism compresses a 20-step task into a validated 5-step efficient path, the overall success rate jumped from 36% to 77%.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/613465API
curl https://kongchang.com/api/v1/knowledge/claims/613465MCP
get_claim(id=613465)