# MiniMax shares SGLang's guide to serving live multimodal AI > Source: > Published: 2026-10-21 19:33:31+00:00 # MiniMax shares SGLang's guide to serving live multimodal AI **The tutorial lays out how one runtime can coordinate voice models, interruptions and persistent sessions; MiniMax's post does not claim the company uses it.** By [RuntimeWire Staff](https://runtimewire.com/author/runtimewire-staff) · Published Primary source: [MiniMax](https://x.com/MiniMax_AI/status/2112988717963178257) ## Why it matters Live multimodal products need infrastructure to coordinate models across a conversation, including interruptions and changing user input. MiniMax's link points to that engineering problem but does not show the company uses SGLang-Omni. On October 21st, [MiniMax](https://www.minimax.io/?ref=runtimewire), founded by [Yan Junjie](https://ir.minimax.io/corporate-information/management?ref=runtimewire), [shared a tutorial and code](https://x.com/MiniMax_AI/status/2112988717963178257?ref=runtimewire) for SGLang-Omni, a runtime designed to coordinate models during ongoing audio conversations. The link points to SGLang's technical work; it is not an announcement that MiniMax adopted the runtime, contributed to it or integrated its models with it.… When you purchase through links in our articles, [we may earn a small commission](https://runtimewire.com/affiliate-monetization-standards). This doesn’t affect our editorial independence.