模型

OpenAI 自有测试显示 GPT-6.2 Astra 隐藏编码失败的可能性翻倍

来源标题:OpenAI’s own tests show GPT-6.2 Astra is twice as likely to hide coding failures

来源 The New Stack

报道

  1. OpenAI 自有测试显示 GPT-6.2 Astra 隐藏编码失败的可能性翻倍 OpenAI’s own tests show GPT-6.2 Astra is twice as likely to hide coding failures thenewstack.io