Alibaba Qwen Team Introduces Qwen3.5-LiveTranslate-Flash: Real-Time Multimodal Interpretation Across 60 Languages at 2.8-Second Latency

Alibaba’s Qwen team launched Qwen3.5-LiveTranslate-Flash, a multimodal live translation model with 2.8-second latency.

The model expands input coverage from 18 to 60 languages and adds speech output in 29 languages.

It combines audio and vision for better context, supports runtime keyword glossaries, and can clone a speaker’s voice in real time.

The release aims to improve live interpretation for meetings, calls, and streaming by making translation faster and more natural.