Open-source Amazon Translate alternatives

LibreTranslate is a machine translation API that you can run entirely on your own server. Unlike hosted translation APIs, it does not depend on proprietary services such as Google or Azure. Its translation engine is the open-source Argos Translate library, and the project is licensed under AGPL-3.0.

It provides a web interface for translating text and files, along with an API documented on the site. The homepage lists languages including Basque, Chinese, English, French, German, Irish, Japanese, Russian and Spanish, among others. The software can work offline and is designed to be set up in a few minutes.

A hosted instance is available at libretranslate.com, with an option to obtain an API key, while developers who prefer control over their data can download the software and run their own API server. It is written in Python, and a Quickstart guide and community resources explain deployment.

Argos Translate is an open-source offline translation library written in Python. It uses OpenNMT for translation and can be used as a Python library, from the command line or through a graphical application. Because models run locally, text can be translated without sending it to an online service. The code is licensed under MIT.

Translation relies on installable language packages, which are zip archives with the .argosmodel extension that contain the data for a language pair. A package index with pre-trained models is maintained in a separate repository. When no direct model exists, Argos Translate can pivot through an intermediate language: with Spanish to English and English to French installed, Spanish to French works too, at some cost in quality.

Supported languages include Arabic, Chinese, Czech, Dutch, French, German, Hindi, Japanese, Korean, Russian, Spanish, Turkish and Ukrainian, among others, and users can request more. The library is on PyPI and installs with pip. LibreTranslate, a separate web app and API, is built on top of Argos Translate, which is how many people encounter it.