The Qwen‑Omni‑realtime model seems to offer very good cost‑performance. I wonder if there are plans to support it via a plugin for LiveKit (or LiveKit Agents) in the near future?

could just make an openai compatible realtime api, and use the openai plugin

Is there any sample or reference code available for this approach?

https://help.aliyun.com/zh/model-studio/realtime#2a9d71e8e87rn could follow to map openai events to qwen's event, and provide your own websocket server