Argos is an open-source machine translation tool available offline

By /

Today, with the popularity of various online translation tools (Google Translate, DeepL, ChatGPT), we are almost accustomed to “having to be connected to translate”.
But in some scenarios, you may encounter limitations:

  • Handle privacy-sensitive documents and cannot be uploaded to the cloud
  • Deploy on servers without an extranet
  • For batch translation, API costs are too high
  • You want to make your own application, but you rely on external services

That’s when it’s important to have a truly offline, open-source, and scalable translation tool.

Argos Translate is one such project.

1. What is Argos Translate?

Argos Translate is a neural network-based machine translation system that operates completely offline.
It is based on the OpenNMT framework, can execute translation models on the CPU, supports multilingual package switching, and provides CLI, GUI, and Python API.

Summary in one sentence:

An open-source Google Translate that runs locally on your computer.

Its most attractive features are:

  • Completely offline, data security
  • Open source can be modified
  • Lightweight, it can run on a normal computer
  • Easy to integrate into your project (Python / Scripting / GUI)

2. How do I install Argos Translate?

Argos supports Windows / macOS / Linux and you can install it your way.

With pip, if you already have a Python environment:

pip install argostranslate

3. Install the language pack

Argos’ translation effect is completely dependent on the language pack.
The official offers dozens of language directions, just choose the one you need.

For example: English → Chinese

argos-translate --install-package translate-en_zh

Or download it in Python:

import argostranslate.package
import argostranslate.translate

# 下载语言包
argostranslate.package.update_package_index()
packages = argostranslate.package.get_available_packages()

# 安装 en→zh 包
package_to_install = next(p for p in packages if p.from_code == "en" and p.to_code == "zh")
package_to_install.install()

Once done, your local translation is ready.

4. Command line translation

Easiest to use:

argos-translate --from-lang en --to-lang zh "Hello world"

Output:

你好世界

It is also possible to translate documents:

argos-translate --from-lang en --to-lang zh input.txt output.txt

5. Desktop GUI translation

Argos officially provides a simple graphical interface (PyQt).

Once installed, it can be used like regular software:

  • On the left, enter the text in the source language
  • Automatically generate target language translations on the right
  • Switchable language packs
  • Offline operation is supported

Suitable for users who don’t write code.

6. Python API integration

This is the most important part if you want to embed translation into your own application.

The simplest call:

from argostranslate import translate

result = translate.translate("Hello, how are you?", "en", "zh")
print(result)

Example of batch translation

sentences = ["Hello world", "This is a test.", "How are you?"]

translated = [translate.translate(s, "en", "zh") for s in sentences]

print(translated)

7. Train your own translation model?

Argos supports custom model training (based on OpenNMT).

You can use:

  • Own parallel corpus
  • Domain Dictionary
  • Industry-specific data (medical, legal, IT)

This allows for more professional translation results rather than using generic models.

The training process is roughly as follows:

  1. Preparing the corpus (src.txt + tgt.txt)
  2. Train the model using OpenNMT-py
  3. Packaged into the Argos language package format
  4. Install this language pack locally

8. Summary

Argos Translate is a very useful open source project that is especially suitable for:

  • Users who need offline translation
  • Developers who don’t want to rely on third-party APIs
  • Businesses with privacy needs
  • Teams that want to customize industry models
  • Developers who want to be local translation tools, automatic summarization systems, and subtitle automation tools

Simple, open-source, and natively available is its greatest value.

Github：https://github.com/argosopentech/argos-translate
Tubing:

Scroll to Top