GPT-6 Astra 物理基准反超专家评分

Astra 在物理基准上坚称专家答案有误，团队复核后确认它是对的，分数从 80% 修正到 98%。

Freddy Vega

@freddier16 天前收录原推发布于 2026.10.15

Freddy Vega @freddier原推发布于 2026.10.15

GPT-6 Astra got +80% on a Physics benchmark. It insisted the expert results were wrong and its calculations were right. A team checked. And yes. GPT-6 Astra was right. New score: 98%

这条是引用着另一条写的：x.com/pratikrath15/status/2110225213401206892

x.com/freddier/status/2110443503516057682采集于 2026.10.15 10:51

在 X 上打开原推