Скачать или смотреть SGLang Office Hour Recap: Deploying GLM-5 with Modal — Feb 11, 2026

In this video, Charles Frye (Modal) walk through deploying GLM-5 end-to-end — from zero to a fully running endpoint on Modal with SGLang. The session covers setup, configuration, performance tuning, and a live Q&A with Charles and Liangsheng Yin (SGLang) on real-world serving tradeoffs for big MoE models.

Timeline & Highlights:

0:00 Introduction & GLM-5 Overview on OpenCode - What GLM-5 is, benchmark performance, and why it matters for open-source AI.
17:37 Deploying GLM-5 on Modal with SGLang Live walkthrough
31:36 Q&A : Future Roadmap / Discussion & Use Cases - Discussion on real-world use case of GLM-5. Discussed future roadmap of speculative decoding improvements, FA4 support on Blackwell, and more.

📑 Slides: https://docs.google.com/presentation/...

💬 Join the community: Follow us on LinkedIn and X for the latest updates. Thanks for watching!
LinkedIn:   / sgl-project  
X: https://x.com/lmsysorg