Model Releases

Tencent open-sources Hy-MT2 multilingual translation model

Original title:🚀 Introduce Hy-MT2: New Open-Source Multilingual Translation Model We proudly launch our new Hy-MT...

AI summary

Tencent has officially open-sourced Hy-MT2, a multilingual translation model supporting seamless translation among 33 languages. Its 7B and 30B-A3B versions achieve state-of-the-art translation performance among open-source models, surpassing many models dozens of times larger in parameter count. More notably, the 1.8B lightweight version outperforms mainstream commercial APIs such as Microsoft's, and with Tencent's AngelSlim 1.25-bit extreme quantization technology it requires only 440MB of storage to run locally on mainstream phone chips, with inference 1.5 times faster than the previous generation, significantly lowering the barrier to deploying high-quality AI translation.

Key Takeaways

  • Tencent has officially open-sourced Hy-MT2,a multilingual translation model supporting seamless…
  • Its 7B and 30B-A3B versions achieve state-of-the-art translation performance among open-source…
  • More notably,the 1.
  • 8B lightweight version outperforms mainstream commercial APIs such as Microsoft's,and with…
  • 25-bit extreme quantization technology it requires only 440MB of storage to run locally on…

Quick Facts

Published
Updated
Source
X:腾讯混元 (@TencentHunyuan)
Category
Model Releases
AI score
74
Read the original Read the full article at the original source
Why it matters

虽然翻译领域不算最热,腾讯这个1.8B开源模型用1.25位量化直接跑在手机上,效果还超微软商业API,做本地化翻译工具的人值得关注。

Related entities

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: Tencent open-sources Hy-MT2 multilingual translation model. May 21, 2026. https://new.nat6.net/en/news/cmpf9b1ua04fzsljw1eu965w5/

View all →