Expired60% confidenceFactExact time
According to the OSWorld benchmark, the most advanced Computer Use models still achieve less than 40% task success rate
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
7/2/2026
First Seen
Valid until: 9/30/2026(expired)
Sources
Agent Tool Selection Guide: A Comprehensive Comparison of API, CLI, MCP, Browser Use, and Computer Use
bilibili慢炖AI6/5/2026
Related Claims
UnverifiedAnthropic 官方承认 Computer Use 在复杂任务上的成功率仅约 60-70%74% similarUnverifiedThe current best models typically achieve task completion rates of 20%–40% on benchmarks such as WebArena and Mind2Web.71% similarUnverified研究表明将复杂需求拆解为多个小任务比一次性提交完整需求的成功率高出40%以上68% similarUnverified据公开评测现有桌面Agent在复杂工作流中的成功率普遍低于50%67% similarUnverified据统计全球个人计算设备的平均CPU利用率不到20%67% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/46722API
curl https://kongchang.com/api/v1/knowledge/claims/46722MCP
get_claim(id=46722)