An open-source offline translation library and toolkit written in Python, supporting over 30 languages via downloadable language models.
Argos Translate is an open-source machine translation library written in Python that enables offline translation between over 30 languages. It uses downloadable language model packages and can pivot through intermediate languages to translate between language pairs without direct models. The project also powers LibreTranslate, a self-hostable API and web application for translation services.
Developers and organizations needing offline, privacy-focused translation capabilities, including those building localized applications, translation tools, or self-hosted translation services.
It offers a completely offline, open-source alternative to cloud-based translation APIs, with support for a wide range of languages through extensible model packages and the flexibility to be used as a library, CLI, or GUI.
Open-source offline translation library written in Python
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
All translations occur locally after model installation, ensuring no data leaves the device, which is ideal for sensitive or regulated environments, as highlighted in the key features.
Offers Python library, CLI, and GUI interfaces, making it versatile for developers, sysadmins, and end-users, with separate installs for different workflows from the README.
Automatically chains translations through intermediate languages to cover more pairs, though with quality loss, enabling broader language support without direct models, as explained in the README.
Supports CUDA via environment variables for faster translations on Nvidia GPUs, leveraging CTranslate2 for improved batch processing, as detailed in the GPU acceleration section.
Translation accuracy can be lower, especially when pivoting, and models may not match commercial services like Google Translate, as admitted in the README regarding quality trade-offs.
Requires downloading and managing separate .argosmodel files for each language pair, which can consume gigabytes of storage and involve manual installation steps, adding overhead.
Primary support is through community forums and GitHub issues, rather than comprehensive docs, which can slow troubleshooting compared to paid or larger open-source projects.