Unverified60% confidenceBenchmarkExact time

GPT-4和Claude 3在HumanEval代码基准测试上的pass@1通过率超过85%

2

Sources

60%

Confidence

Long-term

Relevance

7/12/2026

First Seen

Sources

Base44实战：Claude Code如何支撑1到80人快速扩张

bilibiliGelai_AI7/5/2026

Related Claims

VerifiedGPT-4、Claude 3.5、DeepSeek等模型在HumanEval编程基准测试上的通过率已超过80%83% similar Unverified早期GPT-3的HumanEval Pass@1不足10%，而GPT-4o、Claude 3.5 Sonnet等顶尖模型已稳定超过85%81% similar UnverifiedGPT-4和Claude等前沿模型在代码生成任务中的首次通过率（Pass@1）通常在60%-80%之间，意味着每5次修改中可能有1-2次存在问题77% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/489873

API

curl https://kongchang.com/api/v1/knowledge/claims/489873

MCP

get_claim(id=489873)