OpenAI says planned GPT-6.1 is too insecure to release

Ars Technica作者： Kyle Orland 2026年9月29日查看原文 ↗

Similar performance, security tradeoffs also seen in current public models.