Tencent open-sources Hy-MT1.5-1.8B-1.25bit translation model, 440MB size supports offline phone use
Original title:We're open-sourcing Hy-MT1.5-1.8B-1.25bit — a 440MB translation model that runs fully offline on you...
AI summary
Tencent has open-sourced the Hy-MT1.5-1.8B-1.25bit translation model. It has 1.8 billion parameters and is only 440MB after quantization, allowing it to run fully offline on phones. The model supports 33 languages, 5 dialects, and 1,056 translation directions, including minority languages such as Tibetan and Mongolian. In standard tests, its performance is comparable to commercial translation APIs and a 235-billion-parameter large model. By quantizing to 1.25 bits, memory usage drops sharply from 3.3GB in FP16, with a 25% smaller size and about 10% faster speed than the previous 1.67-bit method, with no loss in accuracy. The model has won 30 first-place awards in international machine translation competitions and has been deployed in multiple Tencent products.
Key Takeaways
- Tencent has open-sourced the Hy-MT1.
- 25bit translation model.
- It has 1.
- 8 billion parameters and is only 440MB after quantization,allowing it to run fully offline on…
- The model supports 33 languages,5 dialects,and 1,056 translation directions,including minority…
Quick Facts
- Published
- Updated
- Source
- X:腾讯混元 (@TencentHunyuan)
- Category
- Model Releases
- AI score
- 67
440MB的模型能在手机上跑33种语言翻译,还宣称比谷歌翻译强,这个量化技术让离线翻译不再是‘能看不能用’,出差党可以试试看。
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.
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-MT1.5-1.8B-1.25bit translation model, 440MB size supports offline phone use. April 29, 2026. https://new.nat6.net/en/news/cmok52zii027oslz3yvh4kwyv/