原帖引用 · 原文
Tibo@thsottiaux
@TokenGremlin Interesting
I have no idea what OpenAI did with GPT-6.1 Terra, but this thing runs like it's in Ultrafast mode when you enable Fast. And even at normal speed, it's ridiculously fast.
I genuinely wasn't expecting this. It caught me completely off guard, and I think OpenAI has pulled off some kind of engineering miracle with this model's inference speed.
OpenAI, please bring this level of optimization to every model going forward.
The downside? Some users still can't access it. I'm already opening a support ticket to get the issue investigated.
在 X 查看这条帖子 ↗
· UTC+8