EasyNMT is an open-source Python package that makes state-of-the-art neural machine translation for more than 100 languages usable in a few lines of code. Built by UKPLab, the team behind Sentence Transformers, it has earned about 1,300 stars on GitHub and is released under the Apache-2.0 license.
Neural translation models are powerful but fragmented: every model family has its own tokenizer, inputs and quirks. EasyNMT wraps leading open models, including OPUS-MT, M2M-100 and mBART, behind one consistent interface. You load a model, pass sentences or whole documents, and get translations back, with automatic language detection included. The package handles long texts by translating document structure sensibly instead of chopping blindly at token limits, and it can run on CPU or GPU. For teams that prefer not to run models themselves, the same simple API is what made EasyNMT a popular quick-start route into multilingual NLP.
Models are downloaded automatically from the Hugging Face hub on first use and cached locally, and a Docker-based server is provided so teams can expose translation as an internal HTTP service without writing glue code. Because EasyNMT shares authors and design philosophy with Sentence Transformers, it follows the same pattern: sensible defaults, a tiny API surface, and an easy path from a notebook experiment to a small production service.
EasyNMT is completely free and open source under the Apache-2.0 license. You only need your own compute to run the underlying translation models.
EasyNMT is for Python developers and NLP teams who need reliable multilingual translation without building pipelines around each model family, from content localization prototypes to research experiments.