FREEDOM251

Freedom251 › Translation Software

EasyNMT

Paid APILinuxSelf-hosted

Visit EasyNMT Alternatives

EasyNMT's own site

What EasyNMT is

EasyNMT is a free, open-source Python package for neural machine translation across more than 100 languages. It translates sentences, documents, and sentence batches, and can detect source languages automatically with fastText support described for more than 170 languages. The package wraps Opus-MT, mBART50, and M2M-100 models; Opus-MT is described as supporting more than 1,200 translation directions across over 150 languages. Docker images expose REST endpoints for translation, language detection, and supported-language queries. EasyNMT supports multi-GPU and multi-process translation, and its documented integrations include PyTorch, fastText, Docker, and Google Colab. Installation uses `pip install -U easynmt`. Opus-MT inputs are limited to 512 word pieces and M2M inputs to 1,024, with longer documents split into sentences. EasyNMT runs through an API, on Linux, or in a self-hosted setup. It is Apache-2.0 licensed, aimed at science and research, and described as experimental software.

Who it suits

EasyNMT suits developers and researchers who want open-source translation in Python, through an API, or in self-hosted Linux and Docker environments. It is positioned for experimental science and research use.

What it does well

  • Translates sentences, documents, and batches
  • Supports more than 100 languages
  • Automatic language detection with fastText
  • Docker REST endpoints available
  • Multi-GPU and multi-process support

What to know first

  • Input lengths are limited by model
  • Longer documents are split into sentences
  • 12-billion-parameter M2M model is unsupported
  • Project is described as experimental

Verdict

EasyNMT offers a flexible, open-source route to neural translation with local, API, and Docker deployment options. Its model input limits and experimental positioning matter for production planning.

Facts from EasyNMT's own pages

API access
Yes · github.com, 22 Sept 2026
API endpoints
The Docker documentation lists translation, language detection, and supported-language endpoints. · github.com, 13 Oct 2026
Document limit
The README states input limits of 512 word pieces for Opus-MT and 1024 for M2M models, with automatic sentence splitting for longer documents. · github.com, 13 Oct 2026
Document translation
Yes · github.com, 22 Sept 2026
Installation
The README gives `pip install -U easynmt` as the installation command. · github.com, 13 Oct 2026
Integrations
The README describes PyTorch-based models, fastText language detection, Docker images, and a Google Colab hosting example. · github.com, 12 Oct 2026
Intended audience
The repository describes the software as experimental and intended to encourage future research. · github.com, 13 Oct 2026
Language coverage
The README says Opus-MT supports 1200+ translation directions across 150+ languages. · github.com, 13 Oct 2026
Language detection
It can automatically detect source languages using fastText, with the README citing support for more than 170 languages. · github.com, 12 Oct 2026
License
The repository lists the Apache-2.0 license. · github.com, 13 Oct 2026
License and security
The repository is published under the Apache-2.0 license; the opened project pages do not state security certifications or compliance claims. · github.com, 12 Oct 2026
Local models
The Docker documentation says model files are stored at `/cache/`, which can be mounted on the host for reuse. · github.com, 13 Oct 2026
Model availability
The README says the 12-billion-parameter M2M model is not supported and lists M2M-100 models with 418 million and 1.2 billion parameters. · github.com, 12 Oct 2026
Model limits
The README states input limits of 512 word pieces for Opus-MT and 1,024 for M2M models; longer documents are split into sentences for translation. · github.com, 12 Oct 2026
Models
Available model wrappers include Opus-MT, mBART50, and M2M_100. · github.com, 13 Oct 2026
Parallel processing
EasyNMT lists multi-GPU and multi-process translation as features. · github.com, 13 Oct 2026
Product
EasyNMT is a Python package for neural machine translation across more than 100 languages. · github.com, 13 Oct 2026
Release
PyPI lists EasyNMT version 2.0.2, released May 27, 2022. · pypi.org, 12 Oct 2026
REST API
The project provides Docker images that expose EasyNMT as a REST API. · github.com, 13 Oct 2026
Support
The author invites users to email or report an issue if something is broken or they have questions. · github.com, 13 Oct 2026
Translation
The project README describes sentence and document translation and automatic language detection for 170+ languages. · github.com, 13 Oct 2026

Where EasyNMT stands on Freedom251

Look at these too