Expired50% confidenceBenchmarkExact time
The strongest AI Agent achieved a task completion rate of only approximately 20% on OSWorld 2.0.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/10/2026
First Seen
Valid until: 10/8/2026(expired)
Sources
OSWorld 2.0测评：最强AI Agent完成率仅20%，长流程是致命短板
bilibili熊二等兵7/1/2026
Related Claims
Unverified来自上海 AI Lab 等机构的研究表明，CodeAct Agent 在多步骤任务基准测试中的成功率比传统 JSON 指令 Agent 高出约 20%67% similarUnverifiedCurrently, top AI Agents (such as Devin and SWE-agent) achieve a resolution rate of approximately 40–55% on SWE-bench.67% similarUnverifiedAI在OSWorld 2.0任务执行过程中仅有不到7%的步骤用于检测和修复自身错误63% similarUnverifiedAccording to an AI engineer's account, 20% of their daily work is AI and the remaining 80% is pure software engineering.63% similarUnverifiedAn AI Agent with core capabilities can be built from scratch using only 200 lines of Python code.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/421496API
curl https://kongchang.com/api/v1/knowledge/claims/421496MCP
get_claim(id=421496)