AlibabaQwen Audio
Qwen Audio 3.1 Realtime
Qwen Audio 3.1 Realtime supports full-duplex speech and text over WebSocket, WebRTC or AOQ. Plus and Flash share acoustic, semantic and manual turn controls.
Practical use
Practical guidance. Use semantic turn detection where pauses do not mark the end of a sentence. Keep retrieved facts separate from completed actions in spoken answers. Check that an interruption does not cause a cancelled tool request to run again.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API |
| Model identifiers | qwen-audio-3.1-realtime-plusqwen-audio-3.1-realtime-flash |
| Languages | 14 languages plus listed Chinese varieties.Qwen Audio real-time guide |
| History | At most 50 audio turns or 300 seconds of retained audio.Qwen Audio real-time guide |
| Flash distinction | Flash can use web search and function calling together; Plus cannot.Qwen Audio real-time guide |
Related models
- Qwen Audio 3.1 TTS
Alibaba · Qwen Audio
Text-to-speech family with Plus for detailed voice performance and Flash for low-latency interaction.
Launch date not confirmedAPI - Qwen Audio 3.1 TTS Next
Alibaba · Qwen Audio
Chinese and English audio generation model that combines speech, sound effects and ambient audio.
Launch date not confirmedAPI