Alibaba's Qwen Releases Live Translation Model with Reduced Latency
By AI123
On September 19, 2026, Alibaba's Qwen officially released Qwen3.8-LiveTranslate, a new simultaneous interpretation model. The model uses an Interleave architecture to restructure real-time interpretation, improving accuracy, fluency, and conciseness, while reducing the average per-character latency (LAAL) from 2.8 seconds to 2.3 seconds.
Explore topicsAlibabaModelsQwen

