Unverified60% confidenceFactTime unknown
在HumanEval等代码能力基准测试中，顶级大语言模型的通过率已超过90%
2
Sources
60%
Confidence
Medium-term (~90 days)
Relevance
5/31/2026
First Seen
Valid until: 8/29/2026
Sources
Vibe Coding入门：产品经理零基础氛围编程指南
bilibiliAgent产品
Related Entities
Related Claims
Unverified基准过拟合是当前大模型评估的核心问题，部分模型在HumanEval上达到90%以上通过率但在真实代码库中表现远不如预期77% similarUnverifiedHumanEval等标准化代码基准测试显示，顶级模型的pass@1得分已超过90%77% similarUnverified2025年的顶尖模型在HumanEval基准测试中的通过率已突破90%76% similarUnverified顶尖大语言模型在MMLU上的得分已普遍突破90%，导致该测试区分度正在下降75% similarUnverified发帖者指出编码任务中缓存命中率通常高达90%以上73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/4734API
curl https://kongchang.com/api/v1/knowledge/claims/4734MCP
get_claim(id=4734)