Skip to content
Trending storyDeveloping

LMSYS Org adds full-duplex real-time interaction to SGLang-Omni

1 article1 sourcesince Oct 20Last article 2h ago ·

Overview

AISummary of 1 article

LMSYS Org says SGLang-Omni now supports real-time interaction through one runtime that covers native full-duplex models, turn-based multimodal models, and cascaded systems.

In single-session tests on one H200, the Qwen3.8-27B cascade with thinking off averaged 0.17 seconds stop latency and 1.32 seconds response latency, according to the company's post on X. Demos and a blog post are available at sglang.io.

Written by AI from the articles below · updated Oct 20, 3:46 PM ET

Check the sources:

Latest developmentOct 20, 3:22 PM ET
SGLang-Omni adds full-duplex real-time interaction across three model types

Article timeline

The articles in this story. Times are ET.

Oct 20
  1. LMSYS OrgOfficial
    SGLang-Omni adds full-duplex real-time interaction across three model types

    AILMSYS Org says SGLang-Omni now supports real-time interaction through a unified runtime covering native full-duplex models, turn-based multimodal models, and cascaded systems. In single-session tests on one H200, the Qwen3.8-27B cascade with thinking off averaged 0.17s stop latency and 1.32s response latency. Demos and a blog post are available at sglang.io.

Heat Trend

Not enough continuous observations to show a trend yet.