Tencent’s Hy‑MT2 is an open‑source multilingual translation model family with 1.8B, 7B, and MoE‑based 30B‑A3B variants supporting translation across 33 languages; the smallest version can run offline on phones at 440... The models are designed specifically for translation rather than general chat, improving instruct...
How does Tencent’s newly open‑sourced Hy‑MT2 multilingual translation model family (Hy‑MT2‑1.8B, 7B, and 30B‑A3B) work, what languages doesTencent’s Hy‑MT2 family includes lightweight mobile models and larger MoE systems for high‑quality multilingual translation.
AI 提示
Create a landscape editorial hero image for this Studio Global article: How does Tencent’s newly open‑sourced Hy‑MT2 multilingual translation model family (Hy‑MT2‑1.8B, 7B, and 30B‑A3B) work, what languages does. Article summary: Tencent’s Hy‑MT2 is a new family of specialized multilingual translation models rather than a general chatbot repurposed for translation. It comes in 1.8B, 7B, and 30B‑A3B sizes, supports 33 languages, and the largest mo. Topic tags: general, academic, general web. Reference image context from search candidates: Reference image 1: visual subject "# Tencent Researchers Release Tencent HY-MT1.5: A New Translation Models Featuring 1.8B and 7B Models Designed for Seamless on-Device and Cloud Deployment. Tencent Hunyuan research" source context "Tencent Researchers Release Tencent HY-MT1.5 - MarkTechPost" Reference image 2: visual subject "It is about proving that a
openai.com
Tencent has open‑sourced Hy‑MT2, a new family of multilingual machine‑translation models designed specifically for translation tasks rather than general chat. The release includes three variants—Hy‑MT2‑1.8B, Hy‑MT2‑7B, and Hy‑MT2‑30B‑A3B—covering use cases from on‑device translation to high‑quality professional translation systems. All models support translation among 33 languages, and the smallest model can run locally on smartphones thanks to aggressive compression techniques.
The Hy‑MT2 Model Family
Hy‑MT2 is structured as a tiered model lineup optimized for different deployment environments:
Hy‑MT2‑1.8B – A lightweight model designed for edge devices and mobile hardware.
Hy‑MT2‑7B – A balanced dense model targeting strong translation quality with manageable compute requirements.
Hy‑MT2‑30B‑A3B – A large Mixture‑of‑Experts (MoE) model intended for top‑tier translation quality in server or cloud environments.
All three models share the same core design goal: handling complex real‑world translation tasks while supporting translation instructions written in multiple languages.
The largest model uses an MoE architecture, which activates only a subset of expert networks during inference. This increases effective capacity while keeping compute costs lower than a fully dense model of comparable size.
Studio Global AI
Search, cite, and publish your own answer
Use this topic as a starting point for a fresh source-backed answer, then compare citations before you share it.
Tencent’s Hy‑MT2 is an open‑source multilingual translation model family with 1.8B, 7B, and MoE‑based 30B‑A3B variants supporting translation across 33 languages; the smallest version can run offline on phones at 440...
首先要验证的关键点是什么？
Tencent’s Hy‑MT2 is an open‑source multilingual translation model family with 1.8B, 7B, and MoE‑based 30B‑A3B variants supporting translation across 33 languages; the smallest version can run offline on phones at 440... The models are designed specifically for translation rather than general chat, improving instruction‑following, domain translation, and real‑world performance compared with earlier Hunyuan translation models.[11][12]
接下来在实践中我应该做什么？
Developers can access weights and code through GitHub, Hugging Face, and ModelScope, while consumers can use the technology via Tencent’s “Hy Translation” WeChat mini‑program and upcoming mobile apps.[3][11]
Hy‑MT2 supports bidirectional translation across 33 languages.
In addition, Tencent reports that the system also supports translation involving five Chinese ethnic languages or dialect variants, expanding its usability in multilingual regions within China.
Although public announcements confirm the 33‑language coverage, the complete official language list is not fully enumerated in the available summaries of the paper. What is clear is that the system focuses strongly on Chinese‑centric multilingual translation scenarios alongside widely used global languages.
A Translation Model That Can Run on a Phone
One of the most notable features of the release is the extremely small deployment footprint of the 1.8B model.
Tencent uses AngelSlim 1.25‑bit quantization, which compresses the model to roughly 440 MB while preserving practical translation quality.
This allows:
On‑device translation on common smartphone chips
Offline translation without internet connectivity
Faster inference compared with previous versions
According to Tencent’s reports, the quantized model also runs about 1.5× faster than the earlier Hy‑MT1.5 generation while maintaining similar deployment size.
Benchmark Performance and Comparisons
Tencent reports strong results for Hy‑MT2 across multiple translation benchmarks and real‑world scenarios.
Key claims from the technical report and release coverage include:
7B and 30B‑A3B models achieve state‑of‑the‑art performance among open translation models on several evaluations.
The models perform strongly on benchmarks such as FLORES‑200, DomainMTBench, and IFMTBench, covering general, domain‑specific, and instruction‑following translation tasks.
The 1.8B model reportedly outperforms some mainstream commercial translation APIs in aggregated benchmarks despite its much smaller size.
In certain tests, the Hy‑MT2 family approaches the performance of leading closed systems like Gemini‑class translation models on FLORES‑200 averages.
Because most benchmark claims come from Tencent’s own report and release materials, independent third‑party evaluations are still limited. However, the reported results suggest Hy‑MT2 is particularly competitive among open translation models, especially relative to its parameter size.
Improvements Over Earlier Hunyuan Translation Models
Hy‑MT2 builds on Tencent’s earlier Hy‑MT1.5 translation models and introduces several improvements.
Major upgrades include:
Better instruction following
The models are optimized to interpret translation instructions such as formatting, terminology control, style changes, or structured translation tasks.
Stronger domain translation
Evaluations emphasize improvements in specialized fields such as finance, education, and other professional contexts.
Improved real‑world translation quality
The training pipeline includes large multilingual datasets and post‑training methods like distillation and reinforcement techniques to improve translation outputs in practical scenarios.
More flexible deployment options
The model family now spans cloud‑scale MoE models, balanced mid‑size models, and ultra‑lightweight edge deployments.
How Developers Can Access Hy‑MT2
Tencent has released the models and code publicly for developers.
Available access points include:
GitHub repositories for code and model resources
Hugging Face model hosting
ModelScope distribution for the Chinese developer ecosystem
The models are designed to run across multiple hardware platforms including ARM and common server architectures.
The lightweight variant in particular targets local deployment scenarios, including edge devices and mobile hardware.
How Consumers Can Use Hy‑MT2
Tencent has also launched consumer products powered by the new models.
The first public interface is the “Tencent Hy Translation” (腾讯Hy翻译) WeChat mini‑program, which allows users to translate text and speech using the Hy‑MT2 system.
Key capabilities include:
Voice input translation
Custom translation style or instructions
Online translation using larger models
Offline translation using locally downloaded models
Tencent has also indicated that dedicated iOS and Android apps are planned, with support for local on‑device inference.
Why Hy‑MT2 Matters
Hy‑MT2 illustrates a broader trend in AI translation: moving from large, general‑purpose models toward specialized multilingual systems optimized for efficiency and deployment flexibility.
By combining:
MoE scaling for high‑quality translation
compact models for edge devices
multilingual instruction following
Tencent is positioning Hy‑MT2 as both a developer‑friendly open translation stack and a practical on‑device translation engine.
If the reported benchmarks hold up under independent evaluation, Hy‑MT2 could become one of the most capable open‑source translation model families—especially for multilingual and mobile translation scenarios.
Comments
0 comments