Alibaba Qwen Team Releases Qwen3.8-LiveTranslate: A Real-Time Interpretation Model That Cuts Average Lag to 2.3 Seconds Across 60 Languages
Alibaba's Qwen team launched Qwen3.8-LiveTranslate, a real-time speech translation model cutting average lag from 2.8 to 2.3 seconds across 60 languages.
Alibaba's Qwen team released Qwen3.8-LiveTranslate, a real-time simultaneous interpretation model built on the Qwen-Omni stack with a new Interleave architecture. It reduces Length-Adaptive Average Lagging (LAAL) from 2.8 seconds to 2.3 seconds, roughly an 18% cut, and adds real-time speaker diarization, synchronized bilingual display, and long-context disambiguation. The model understands 60 languages and speaks 29, accepting audio and optional images as input. It is API-only via Alibaba Cloud Model Studio and QwenCloud as qwen3.8-livetranslate-flash-realtime over WebSocket, with a 53,248-token context window and Singapore pricing of $7.50 per 1M audio input tokens.
- New Interleave architecture cuts LAAL average lag from 2.8s to 2.3s, roughly an 18% reduction.
- Understands 60 languages, speaks 29, with real-time speaker diarization and voice cloning.
- Adds synchronized bilingual display and long-context disambiguation for names and terminology.
- Available via WebSocket API on Alibaba Cloud Model Studio and QwenCloud; 53,248-token context window.
- Supports up to 1,000 hotwords and visual cues from images sent at a maximum of 2 per second.
- Tags
- alibabamultimodalqwenqwen3.8-livetranslatereal-time-apisimultaneous-interpretationspeech-translationvoice-cloning
- Organizations
- Alibaba
From the original article
3 min read at marktechpost.com
Qwen has released Qwen3. 8-LiveTranslate, a real-time simultaneous interpretation model built on a new Interleave architecture. It cuts average lagging (LAAL) from 2.
By Asif Razzaq · marktechpost.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.