Hy-MT2-7B-GGUF

Hy-MT2-7B-GGUF

Translates text among 33 languages and follows translation instructions, as a GGUF copy of Tencent's 7B model for llama.cpp. For developers who run translation locally.

Made by tencent on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This is the GGUF version of Hy-MT2-7B, a translation model from Tencent. The Hy-MT2 family covers three sizes and handles translation among 33 languages, and it follows translation instructions given in several languages. The GGUF format is meant for llama.cpp, so you can run the 7B model on your own machine.

Who it suits: It suits developers and teams who want machine translation they host themselves, and people who want to tell the translator how to translate, for example in a given tone or style.

Industry | Who uses it | How they use it | Example
Software | App developers | Add translation to a product without a hosted service | Translate interface strings into another language
Media | Subtitle teams | Translate video subtitles with instructions about style | Convert subtitles while keeping a stated register
Research | Machine translation researchers | Compare an open 7B translator against other open models | Test instruction-following in translation tasks

People who need a tiny download can skip it, since the smaller 1.8B family members and their lower-bit GGUF files exist for that.

Where to use it:

• llama.cpp and tools built on the GGUF format
• Windows, macOS or Linux, where you can download the files with Python
• Translation features inside your own apps and scripts

Use case | Who | What they get back
Translate text between supported languages | Developers | The translated text in the language you name
Follow a translation instruction | Writers and localizers | A translation shaped by your instruction
Run translation offline on your own hardware | Privacy-minded teams | Translations without sending text to a hosted service

How to set it up:

• Python with pip, to download the files
• A GGUF runtime such as llama.cpp
• Around 3.8 GB of disk, the 4-bit estimate for the weights

1. Install the Hugging Face hub client.
2. Download the model files.
3. Point llama.cpp at the downloaded GGUF file.

pip install -U huggingface_hub
hf download tencent/Hy-MT2-7B-GGUF

How to use it: Give the model the text to translate along with a plain instruction about the language pair or the style you want. The instruction examples in the model card show the pattern, and they come in Chinese and English.

Example: You paste a paragraph and ask for a translation into another of the 33 supported languages, adding an instruction about how formal it should sound. The model returns the paragraph in the target language.

For newcomers and experts: This is a translation engine rather than a chatbot. Download the files, load them in llama.cpp, and start by translating one short sentence into a language you can check.

Hy-MT2 is described as a fast-thinking family of 1.8B, 7B and 30B-A3B mixture-of-experts models, with this file being a GGUF build of the 7B one at about 7.5B parameters. The card reports that the 7B and 30B-A3B models beat several larger open-source models in fast-thinking mode. Tencent also released IFMTBench, a benchmark for translation instruction-following.

You give text and an instruction
llama.cpp loads the GGUF file
The 7B model translates
Translated text comes back

Good to know:

• Languages: translation among 33 languages
• Size: about 7.5B parameters and about 3.8 GB at 4-bit, as an estimate
• Family: other sizes and quantizations, including 1.8B and 30B-A3B, are linked from the card
• License: Apache 2.0
• Cost to run: none named
• Data it reaches: the text you give it. The card names no web or account access
• Not for: chat or other tasks, since it is built for translation

On Hugging Face, 13 November 2026

• Parameters: 7.5B
• Weights: about 3.8 GB at 4-bit (estimate)
• Formats: GGUF
• Downloads in the last 30 days: 1,158,882
• Likes: 104
• License: apache-2.0
• Based on: tencent/Hy-MT2-7B
• Tags: gguf, conversational

Links

• Model: https://huggingface.co/tencent/Hy-MT2-7B-GGUF
• Files: https://huggingface.co/tencent/Hy-MT2-7B-GGUF/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download tencent/Hy-MT2-7B-GGUF

Then follow the usage in its model card. Nothing in this kit runs by itself.