# Tencent releases Hy-SimulMT, a real-time Chinese-English speech translation model

> Original title: tencent/Hy-SimulMT-30B-A3B

- Source: Tencent · new models on Hugging Face
- Published: 2026-10-22T07:29:17.000Z
- Bad Influence: https://badinfluence.news/items/tq1f0pxtajlco48z5af901zco
- Original: https://huggingface.co/tencent/Hy-SimulMT-30B-A3B

## Summary

Tencent Hunyuan releases Hy-SimulMT, a real-time simultaneous translation model for bidirectional Chinese-English speech-to-speech translation with real-time voice cloning. On the RealSI long-form benchmark, the model reports a StreamLAAL of about 3 seconds, the lowest among the systems shown, and a GEMBA score above the compared open-source models and four of five commercial APIs. The release includes several model variants on Hugging Face and ModelScope, plus the SimulInfer inference service and SimulTrain training tools on GitHub.

## Why it matters

The source gives benchmark scores, a latency figure, and deployment steps, letting readers judge the real-time translation trade-off and whether to run it locally.