OpenAI 内部用 Astra 重做推理栈，催生 6.1 Sol

Vaibhav Sisinty @VaibhavSisinty原推发布于 2026.10.02

在 X 上打开原推

Here's what's actually wild about what's happening inside openai right now. Tibo explained how gpt-6.1 sol was even possible they gave their smartest engineers astra, and those engineers started doing things previously considered almost impossible. They used astra to rethink the entire inference stack. that's how 6.1 sol is almost as intelligent as astra at a fifth of the price. "we push the frontier and then we make things more affordable and more broadly available." but it's not just models. ultra fast has changed how openai itself ships.

这条是引用着另一条写的：「OpenAI 高管访谈：GPT-6.1 Sol 与 GPT-6 Astra 的区别」