# YottaLabs shares SGLang-Trainium work on new MoE models and inference latency

> Original title: @YottaLabs chief scientist Dong Li shares the team’s work on SGLang-Trainium: adding support for new MoE models and optimizing their infe...

- Source: SGLang
- Published: 2026-11-07T00:28:52.000Z
- Bad Influence: https://badinfluence.news/items/mncrpkbhw8kn2iorch5pkqo84
- Original: https://x.com/sgl_project/status/2118863218915541434

## Summary

YottaLabs chief scientist Dong Li shares the team's work on SGLang-Trainium, adding support for new MoE models. The work also optimizes inference latency for those models on Trainium.

## Post

@YottaLabs chief scientist Dong Li shares the team’s work on SGLang-Trainium: adding support for new MoE models and optimizing their inference latency on Trainium.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWey5vjbAAAwa8p.jpg&mode=thumb&exp=1794268800&sig=c0306013a5e4a25b)