By Tommi Nieminen



OPUS-CAT plugin is the only machine translation plugin that offers completely free, secure, and confidential neural machine translation for over a thousand language pairs.
It also offers a feature that is crucial for professional translators: the possibility to fine-tune the machine translation with bilingual material related to the translation job at hand. Fine-tuning can in many cases have an immense effect on the usability of machine translation.

Note: OPUS-CAT plugin fetches translations from a local MT engine, which should be installed and running on the local computer when the plugin is used. For installation instructions please visit the OPUS-CAT Trados Plugin website to ensure you install and configure your local MT engine correctly. The OPUS-CAT Trados plugin is part of a software ecosystem built around the OPUS corpus (open parallel corpus), which is a large collection of freely available bilingual texts.
Another part of this ecosystem is OPUS-MT, which is a collection of machine translation models trained from the texts in the OPUS corpus. OPUS-MT contains machine translation models for over a thousand language pairs. OPUS-CAT itself is a project that makes it possible to use the OPUS-MT models in CAT tools, such as Trados.

To use OPUS-CAT Trados plugin, first install the OPUS-CAT MT Engine on your local computer. OPUS-CAT MT Engine is an application which makes it possible to use and fine-tune the OPUS-MT models locally without any connection to outside networks.
It is based on Marian NMT, a state-of-the-art neural machine translation framework, known for its speed and efficiency. Thanks to the efficiency of Marian NMT, OPUS-CAT MT Engine works on normal computers without special hardware.

Technical details - Trados Studio 2022


  • Updated to support Trados Studio 2022 SR2.

Checksum: e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855

Release date: 2024-04-26 - Trados Studio 2021

Support website: https://github.com/Helsinki-NLP/OPUS-CAT/issues

Support e-mail:

Shared products:

No related information.

No related information.