Unverified85% confidenceFactTime unknown
在20项复杂任务的测试中,Devin仅成功完成了3项
1
Sources
85%
Confidence
Medium-term (~90 days)
Relevance
5/31/2026
First Seen
Valid until: 8/29/2026
Sources
Devin 2.0深度评测:月费20美元的AI编程代理到底值不值
bilibili攒钱换房车的福叔
Related Entities
Related Claims
UnverifiedDevin在实际使用中20个任务仅成功3次83% similarUnverified在三个递进难度的编码任务测试中,Cursor + GPT-5仅完成了1/3的任务(33%完成率),只完成了Lint错误修复,未能完成静态转动态数据和3D Context Loss修复61% similarUnverifiedEach of the 5 test tasks in the evaluation took approximately 7-10 minutes to complete60% similarUnverified在沙盒物理模拟测试中,Qwen3 27B主动编写测试脚本进行26项断言检查,反复验证直到10/10全部通过60% similarUnverified创建新Skill时应准备10-20条测试提示,包括5条快乐路径场景和5条该不用的场景60% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/3196API
curl https://kongchang.com/api/v1/knowledge/claims/3196MCP
get_claim(id=3196)