Qwen3.8-LiveTranslate: real-time simultaneous interpretation model
Melvin Vivas · X post · 2026-09-19 · Open on X
Topics: Industry Trends & Job Market, LLM Fundamentals · Level: intermediate
Summary
The creator shares Qwen's announcement of Qwen3.8-LiveTranslate, a model for real-time simultaneous interpretation. It uses an Interleave architecture to improve faithfulness, fluency and conciseness, and it cuts average lag across 60 languages.
Key points
- Qwen3.8-LiveTranslate is Qwen's next-generation real-time simultaneous interpretation (speech translation) model.
- It is built on an 'Interleave' architecture.
- Qwen says it improves faithfulness, fluency and conciseness of translations.
- Average lagging (LAAL) dropped from 2.8s to 2.3s.
- It covers 60 languages.
- LAAL (Length-Adaptive Average Lagging) is a standard latency metric for simultaneous translation, useful to know when you compare streaming models.
Resources mentioned
- Qwen3.8-LiveTranslate · tool · qwen.ai · check price
Qwen's real-time simultaneous interpretation model covering 60 languages with about 2.3s average lag.
More in Industry Trends & Job Market
- OpenAI Releases GPT-6 Sol and GPT-6 Luna
- Claude Opus 5.5 Released: Fable 5.1-Level Performance at 40% Lower Cost
- Qwen-Image-2.1 Announced as an Open-Weights Image Model
- Open models now dominate token volume on Vercel AI Gateway
- AuK: Open-Source Model for Generating and Editing Speech
- Qwen3.8-Omni-Flash: Qwen's agentic omni-modal model