Unverified50% confidenceBenchmarkExact time
即使某个新版本发布已过去20周，也只有40%的图像使用最新版本生成
1
Sources
50%
Confidence
Long-term
Relevance
7/20/2026
First Seen
Sources
Related Claims
Unverified研究显示信息不完整的缺陷报告会使开发人员的修复时间平均延长 40% 以上61% similarUnverified在SWE-bench Full版本上，多数模型的通过率仍低于20%59% similarUnverifiedStatistics show that 40% of multi-agent pilot projects fail within six months of going live.58% similarUnverifiedTraditional inference solutions often achieve GPU memory utilization of only 20%–40%.57% similarUnverified2024 年初最先进的模型在 OS World Verified 测试上通常连 20% 都拿不到56% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/569519API
curl https://kongchang.com/api/v1/knowledge/claims/569519MCP
get_claim(id=569519)