跳到正文
Lisan al Gaib· @scaling01 · X·· 3 小时前AI 评分38
AI 导读

想象一下 Astra 或 Fable 跑到 400 tokens/s

正文

Imagine Astra or Fable at 400 tokens/s

引用SemiAnalysis@SemiAnalysis_
BREAKING NEWS🚨: NVIDIA just submitted its first Rubin NVL72 Kimi K3 submission, and it can reach over 850 tok/s/user, even at 2.8 Trillion parameters, with over 3x higher throughput on @vllm_project! Great work by the NVIDIA team, and excited for Rubin Software 1.0 GA coming soon!
在 X 查看被引用的帖子

来源:Lisan al Gaib · x.com