Unverified70% confidenceFactTime unknown
2024年初SWE-Bench最好的系统得分仅在30%左右
1
Sources
70%
Confidence
Long-term
Relevance
6/3/2026
First Seen
Sources
Related Entities
Related Claims
Unverified2024年以前主流模型在SWE-bench上的通过率普遍低于5%，Claude 3.5系列发布后跃升至49%（SWE-bench Verified子集）59% similarUnverifiedCopilot功能与使用是GH-300官方权重最高的领域，约占25%~30%59% similarUnverifiedGrok 4.6's score on Terminal Bench jumped from 15% in the previous generation to 26%.57% similarUnverified评测者认为通义千问3.6 27B是30B级别中对高考数学题表现最好的模型56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/39864API
curl https://kongchang.com/api/v1/knowledge/claims/39864MCP
get_claim(id=39864)