In this video, we demonstrate how to set up translation with ChatGPT in Trados, and how to customize the output with Custom.MT Console.
Custom.MT Console is a machine translation platform integrated with Trados, memoQ, and Smartling. It allows you to create, manage, and use your own custom machine translation (MT) models, and to connect with large language models such as OpenAI GPT-4, GPT-4o, and GPT-4-Turbo for translation and post-editing.
You can customize GPT machine translation output with prompts and glossaries, allowing them to produce translations that align more closely with your organization’s terminology and style. You can also adjust to the reader, and specify their preferred level of text complexity, formal or informal voice, and gender.
In the video below you will find some useful post-editing prompts that you can start your experiments with and advice on how to customize them to better fit your desired outcome:
Frequently Asked Questions (FAQs)
You can write specific instructions as a prompt for ChatGPT or GPT-4 in Custom.MT console. Go to Templates, set the engine to OpenAI, and you will have a prompt engineering box come up. You can set the model’s role, write the style instructions, and add a glossary.
Yes, you can use multiple custom MT models in a single project. You can switch between different models in the Console, and the next segment you translate in Trados will come from the next model.
The quality of your custom MT model is heavily dependent on the level of detail in the prompt. Specify instructions as a bullet point list. You can be vary specifics and target individual errors.
Yes, you can use the custom translation prompts that you created in the Console in memoQ and Smartling. You can also integrate other applications using Custom.MT API.
Yes, every Trados account connected to your Console can benefit from the prompts that you have created.
ChatGPT and GPT-4 support over 90 languages, but the quality of translations varies. It’s safe to assume the main European and Asian languages have stronger support, and smaller languages need to be tested before production use.
Custom.MT is an additional cost of 30 euros per million characters.
To use the models, you may add your OpenAI key. The pricing for OpenAI models can be seen at the company’s official site. As per January 2025, GPT-4o is $2.50 / 1M tokens for input and $10.00 / 1M tokens for output
If you encounter problems with your custom MT model, you can contact Custom.MT support team for assistance at support@custom.mt.
The Custom.MT feature is only available in Trados Studio starting with the 2019 version. If you’re using an older version of Trados Studio, you will need to upgrade to use this feature.
Try GPT in Trados now! Sign in to CustomMT Console for free and upload Trados connector. https://custom.mt/trados-connector/
