gpt-6IMAGE
Astra 一出场就刷满新物理基准 CritPt
作者感叹 OpenAI 的强化学习流水线太强,Astra 直接饱和了这道新物理题,预计 Fable 5.5 也会同样刷满。
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)
@teortaxesTex16 天前收录原推发布于 2026.10.15


Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞) @teortaxesTex原推发布于 2026.10.15
Physics problems. I guess this would have been a nice new CritPt, if Astra didn't saturate it right out of the gate. OpenAI does have a ferociously strong RL pipeline. Fable 5.5 will probably saturate it too.
这条是引用着另一条写的:x.com/pratikrath15/status/2110225213401206892
x.com/teortaxestex/status/2110440866200641847采集于 2026.10.15 10:50