Qwen3.8-LiveTranslate upgrades real-time interpreting with speaker separation, bilingual source-and-translation output and long-context disambiguation. Qwen reports average lag falling from 2.8 to 2.3 seconds while retaining 60 input languages; the API produces speech in 29 languages through a 53,248-token context window. The production model is available over Alibaba Cloud Model Studio's WebSocket Realtime API in Beijing and Singapore.
Qwen3.8-LiveTranslate adds speakers and context to real-time interpreting
Qwen3.8-LiveTranslate upgrades real-time interpreting with speaker separation, bilingual source-and-translation output and long-context disambiguation. Qwen reports average lag falling from 2.8 to 2.3 seconds while retaining 60 input languages; the API produces speech in 29 languages through a 53,248-token context window. The production model is available over Alibaba Cloud Model Studio's WebSocket Realtime API in Beijing and Singapore.
Availability: stable WebSocket Realtime API in Alibaba Cloud Model Studio, Beijing and Singapore
Imported from the existing research archive. Migration preserves the recorded judgment and is not a fresh verification.
What this means for you
For multilingual meetings, support and live events, speaker identity and terminology consistency often matter more than another small generic translation gain. Trial it with overlapping speakers and domain vocabulary, but compare end-to-end latency and translation quality independently because the headline quality claims are vendor-reported.
Evidence & related material
Technical details & scores
| Model Id | qwen3.8-livetranslate-flash-realtime |
|---|---|
| Architecture | Hybrid-MoE Thinker-Talker with interleaved streaming understanding, text and speech |
| Languages | 60 input languages; speech output in 29 languages |
| Latency | vendor-reported LAAL 2.8s to 2.3s |
| Context | 53,248 tokens; maximum input 49,152 and output 4,096 |
| Availability | stable WebSocket Realtime API in Alibaba Cloud Model Studio, Beijing and Singapore |