Unverified50% confidenceBenchmarkExact time
Codex于2021年发布时在HumanEval基准上达到28.8%的pass@1得分，而GPT-4在同一基准上已超过67%
1
Sources
50%
Confidence
Long-term
Relevance
7/11/2026
First Seen
Sources
Connecting DeepSeek V4 in VSCode: Complete Claude Code Plugin Setup Tutorial
bilibili讲AI的薯条学姐7/10/2026
Related Claims
UnverifiedThe original Codex achieved a pass@1 score of 28.8% on the HumanEval benchmark, whereas using GPT-3 directly yielded only 0%.83% similarVerifiedCodex achieved a pass@1 accuracy of 28.8% on the HumanEval benchmark76% similarUnverified初版 Codex 采用约120亿参数规模，在 HumanEval 基准测试中达到 28.8% 的 pass@1 通过率69% similarUnverifiedEarly GPT-4 had a pass rate of only 1.7% on SWE-bench, Claude 3 Opus approximately 4.8%, and it was not until mid-2024 that the 20% mark was surpassed.68% similarUnverifiedThe median output token count per problem-solving attempt for Opus 4.7 is 60,000, while GPT 5.5 requires only 16,000.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/486405API
curl https://kongchang.com/api/v1/knowledge/claims/486405MCP
get_claim(id=486405)