# RadixArk launches SGLang-TorchTPU, a TPU plugin for SGLang

> Original title: SGLang-TorchTPU gives people options to choose the hardware that best fits their workloads and economics. We're excited to see many partn...

- Source: RadixArk
- Published: 2026-10-22T02:03:24.000Z
- Bad Influence: https://badinfluence.news/items/dmnikfjjrxabqnitzq02pi33n
- Original: https://x.com/radixark/status/2113088803170877657

## Summary

RadixArk launches SGLang-TorchTPU, a TPU plugin for SGLang built with Google's TPU team and Meta's PyTorch team. The plugin supports open models including Muse Spark Open, GLM-5.3 Flash, Kimi K3, Ling 3.1 Flash, and MiniMax H3, and the post says Muse Spark Open reaches 770 tok/s decode at bs=1 on InferenceX 8K/1K. The post also says Meta uses it for production inference, and the code is available on GitHub.

## Post

SGLang-TorchTPU gives people options to choose the hardware that best fits their workloads and economics. We're excited to see many partners already running it in production!  

Huge thanks to everyone at Google, Meta, our @sgl_project team, and community who made this happen!  

We're committed to helping make TPU a first-class citizen in the SGLang ecosystem and will keep building toward a more flexible, heterogeneous future for AI infrastructure.

## Quoted post @sgl_project

> Today we launch SGLang-TorchTPU, built with @Google TPU team and Meta @PyTorch team.
> 
> We’re excited to see it powering production inference at Meta!
> 
> SGLang-TorchTPU is a TPU plugin for SGLang that brings:
> 
> - Top open models on TPU: Muse Spark Open, GLM-5.3 Flash, Kimi K3, Ling 3.1 Flash, MiniMax H3
> - SGLang’s production features for large agentic workloads
> - Custom MegaKernel: Muse Spark Open hits 770 tok/s decode at bs=1 on InferenceX 8K/1K
> 
> SGLang-TorchTPU is available now: http://github.com/sgl-project/sglang-torchtpu

https://x.com/sgl_project/status/2112948714453262363