OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Opus Mt

Opus-MT is an open collection of neural machine translation models and service software built around Marian-NMT and training data from OPUS. It provides downloadable pretrained models and includes a Tornado web application for configured language pairs, plus a separate WebSocket service setup.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Opus Mt — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Opus-MT when you want to assemble a translation service from pretrained OPUS-trained models and can configure Marian-NMT and the relevant language-pair services.

Who it is for

  • Developers building translation interfaces or services around pretrained models.
  • Teams selecting models for particular source and target language pairs.
  • Language technology practitioners investigating OPUS-based machine translation resources.

Documented uses

  • Find downloadable pretrained translation models for a language pair.
  • Configure a web application to serve translations for selected language pairs.
  • Provide translation through a web interface, API or the separately documented WebSocket setup.
Before you choose

Requirements & limits

The project is published in a public GitHub repository with an MIT license. The README identifies downloadable public models as CC-BY 4.0; model and software licensing are distinct.

  • The Tornado application's available languages depend on the language pairs configured in its service file.
  • Manual deployment requires Marian MT and its server binary; the README also documents Docker deployment.
  • The README describes most models as using SentencePiece-based segmentation.
  • Training scripts are described as customized for the University of Helsinki and CSC computing environment.
  • Development model variants may cover additional language pairs but are described as more experimental and potentially lower quality.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Launch the web app with Docker

Start the documented Opus-MT web application using Docker.

Before you start

  • Docker environment

Steps

  1. From the project directory, run docker-compose up; alternatively, build the image with docker build . -t opus-mt.

  2. If building the image directly, run docker run -p 8888:8888 opus-mt:latest.

  3. Open localhost:8888 in a browser.

Expected outcome

The Docker instructions expose the web interface on port 8888.

Guide limitations

  • The README does not provide a separate model-selection walkthrough for the Docker route.
  • The documented browser address is localhost:8888.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
Helsinki-NLP/Opus-MT · Official homepage title · retained in its source language
How we review sourcesSuggest a correction