Unverified50% confidenceBenchmarkExact time
OpenAI于2021年发布Codex技术报告，该模型在HumanEval基准测试上的首次通过率达28.8%，经迭代改进后超过72%
1
Sources
50%
Confidence
Long-term
Relevance
7/8/2026
First Seen
Sources
Related Claims
VerifiedHumanEval was proposed by OpenAI in 2021 and evaluates function-level code completion.68% similarUnverified规模定律由OpenAI研究人员于2020年系统量化，后经Chinchilla论文（2022）修正66% similarVerifiedThe earlier Codex API was released by OpenAI in 2021.65% similarUnverifiedOpenAI open-sourced its Evals framework in 2023, categorizing evaluation into three major types: fact-based exact matching, model-based judgment, and human-annotation-based evaluation.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/210563API
curl https://kongchang.com/api/v1/knowledge/claims/210563MCP
get_claim(id=210563)