SGLang 优化 V4.1 Flash：单用户 538→943 tok/s，单卡吞吐 348K→654K

SGLang @sgl_project原推发布于 2026.10.27

在 X 上打开原推

We’ve been working to push both ends of the curve on DeepSeek V4.1 Flash 🚀 • For low latency, our MegaKernel optimizations take interactivity from 538 to 943 tok/s/user. • For high throughput, WideEP and system-wide optimizations take total throughput (cache + new input + output) from 348K to 654K tok/s per GPU. There’s still lots of room for improvement. We’ll share more updates soon.

这条是引用着另一条写的：x.com/semianalysis_/status/2114779348838338719