Crowdin
Localization & translation management platform for apps, websites and docs
Open-Source Projects
Discover top open-source software, updated regularly with real-world adoption signals.
Discover top open-source software, updated regularly with real-world adoption signals.
Offline Python translation library with extensible language models
Argos Translate provides offline machine translation in Python, supporting a wide range of languages via installable .argosmodel packages, with CLI, library, and GUI options, plus automatic pivoting.
Argos Translate is a Python‑based library that delivers offline machine translation. By installing language model packages (files with a extension) you can translate between more than 30 languages without any network connection.
.argosmodelThe library can be used directly in code, through a command‑line interface, or via a separate GUI application. It automatically pivots through intermediate languages when a direct model is unavailable, expanding coverage at the cost of some quality loss. GPU acceleration is available by setting the ARGOS_DEVICE_TYPE environment variable, leveraging CTranslate2 for faster inference. Additional utilities enable HTML and file translation, and the same engine powers the LibreTranslate API and web app.
Install with a single pip install argostranslate command, then manage models with the argospm tool or the Python API. The solution runs on any platform that supports Python, and can be bundled into virtual environments or container images for reproducible deployments.
When teams consider Argos Translate, these hosted platforms usually appear on the same shortlist.
Looking for a hosted option? These are the services engineering teams benchmark against before choosing open source.
Localizing a desktop application
Translate UI strings on the user's machine without external calls
Batch translating documents
Process large sets of files offline, preserving confidentiality
Integrating translation into a chatbot
Provide multilingual responses without latency from external services
Running translation on edge devices
Deploy on Raspberry Pi with GPU acceleration for fast on‑device inference
Use `argospm update` to refresh the index, then `argospm install translate-<src>_<tgt>` or install via the Python API with `package.install_from_path(...)`.
Yes, set `ARGOS_DEVICE_TYPE=cuda` (or `auto`) before running; the library forwards the setting to CTranslate2 for accelerated inference.
Argos Translate automatically pivots through an intermediate language (e.g., source → English → target) using available models.
The core library does not provide a UI, but the separate `argostranslategui` package offers a desktop GUI, and LibreTranslate builds a web API on top of Argos Translate.
Argos Translate is released under the MIT license.
Project at a glance
ActiveLast synced 4 days ago
