Unverified50% confidenceFactExact time
主流AI代理完成OSWorld 2.0每个任务平均需要超过300步，而1.0版本只需30步左右
1
Sources
50%
Confidence
Long-term
Relevance
7/9/2026
First Seen
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在OSWorld 2.0最长的工作流中AI完成率几乎降至0，完成率对任务长度极其敏感72% similarUnverifiedUP主最终使用OpenAI的Codex桌面版处理代码任务，单个任务处理时间约21分钟63% similarUnverifiedIn the Copilot Mode stage (2024), AI handled roughly one-third of the workload with the remaining two-thirds completed by humans62% similarUnverified使用Superpowers后，AI将评论功能需求自动拆解为五个开发任务，并启动子代理模式进行多任务并行开发，约十分钟完成62% similarUnverifiedOSWorld 2.0包含108个长流程真实计算机任务61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/338170API
curl https://kongchang.com/api/v1/knowledge/claims/338170MCP
get_claim(id=338170)