# Modal and SGLang support Kimi K3.1 Pro serving on day one

> Original title: We collaborated on the model bringup with @sgl_project, validating checkpoint accuracy and optimizing the serving path to support Kimi K3...

- Source: Modal
- Published: 2026-11-04T18:04:58.000Z
- Bad Influence: https://badinfluence.news/items/b69emr7jeu53u201oyjq8ovck
- Original: https://x.com/modal/status/2118041832533762424

## Summary

Modal says it collaborated with SGLang on the model bringup for Kimi K3.1 Pro, validating checkpoint accuracy and optimizing the serving path for Day 0 support. SGLang says Kimi K3.1 Pro replaces dense MLA with sparse attention (DSA), which it calls more efficient and easier to serve. SGLang reports up to 33.7K tok/s per GPU on GB300 and 450 tok/s for a single user with DFlash2 and MegaKernel.

## Post

We collaborated on the model bringup with @sgl_project, validating checkpoint accuracy and optimizing the serving path to support Kimi K3.1 Pro on Day 0.
https://x.com/sgl_project/status/2118017145548451860?a=j6Hf-0G7ckk80aoguqo&s=20

## Quoted post @sgl_project

> Kimi K3.1 Pro replaces dense MLA with sparse attention (DSA). It's much more efficient and easier to serve.
> 
> SGLang delivers on Day 0:
> · Up to 33.7K tok/s per GPU (AgentX, GB300)
> · 450 tok/s single user with DFlash2 + MegaKernel
> · NVIDIA Vera Rubin / Blackwell / Hopper, AMD, TPU, Trainium, PPU

https://x.com/sgl_project/status/2118017145548451860