Hunyuan-MT Technical Report
Mao Zheng, Zheng Li, Bingxin Qu, Mingyang Song, Yang Du, Mingrui Sun, Di Wang
2025-09-11
Summary
This paper introduces two new machine translation models, Hunyuan-MT-7B and Hunyuan-MT-Chimera-7B, designed to translate between a large number of languages, with a particular focus on Mandarin Chinese and its related dialects and minority languages.
What's the problem?
Existing machine translation models often struggle with translating between many different languages at once, and they especially have trouble with languages that don't have a lot of available data for training, like many minority languages and dialects. Getting accurate translations for these less common languages is a significant challenge.
What's the solution?
The researchers created Hunyuan-MT-7B, a multilingual translation model trained using a three-step process: first, a broad pre-training phase to learn general language skills, then a focused fine-tuning step to specialize in translation, and finally, reinforcement learning to improve accuracy. They also developed Hunyuan-MT-Chimera-7B, which takes multiple translation suggestions from Hunyuan-MT-7B, generated with slightly different settings, and combines them to produce even better results – similar to how people carefully think through a problem.
Why it matters?
These models represent a significant advancement in machine translation, especially for languages that are often overlooked. They achieve top performance across many language pairs, even outperforming larger models, and are particularly strong in translating to and from Mandarin Chinese and its related languages. This means better communication and access to information for a wider range of people.
Abstract
In this report, we introduce Hunyuan-MT-7B, our first open-source multilingual translation model, which supports bidirectional translation across 33 major languages and places a special emphasis on translation between Mandarin and several ethnic minority languages as well as dialects. Furthermore, to serve and address diverse translation scenarios and enhance model performance at test time, we introduce Hunyuan-MT-Chimera-7B, a translation model inspired by the slow thinking mode. This model integrates multiple outputs generated by the Hunyuan-MT-7B model under varying parameter settings, thereby achieving performance superior to that of conventional slow-thinking models based on Chain-of-Thought (CoT). The development of our models follows a holistic training process specifically engineered for multilingual translation, which begins with general and MT-oriented pre-training to build foundational capabilities, proceeds to Supervised Fine-Tuning (SFT) for task-specific adaptation, and culminates in advanced alignment through Reinforcement Learning (RL) and weak-to-strong RL. Through comprehensive experimentation, we demonstrate that both Hunyuan-MT-7B and Hunyuan-MT-Chimera-7B significantly outperform all translation-specific models of comparable parameter size and most of the SOTA large models, particularly on the task of translation between Mandarin and minority languages as well as dialects. In the WMT2025 shared task (General Machine Translation), our models demonstrate state-of-the-art performance, ranking first in 30 out of 31 language pairs. This result highlights the robustness of our models across a diverse linguistic spectrum, encompassing high-resource languages such as Chinese, English, and Japanese, as well as low-resource languages including Czech, Marathi, Estonian, and Icelandic.