One plugin to access 30+ MT engines, including ChatGPT that can be customized according to your needs.

Custom.MT enables users to plug additional machine translation brands into memoQ.
Your own models built on MarianNMT, Fairseq, and large language models may also be connected to Custom.MT’s platform.













What you can do with Custom.MT?
Using Custom.MT Platform, you can craft your own prompts for GPT-4, receiving tailored translations in memoQ. This allows you to:
Upload your translation memory to Custom.MT Console and choose machine brands for evaluation. Custom.MT will refine and translate using selected MT engines, comparing them to your human translation. Quickly receive BLEU, COMET, and WER scores to determine the best MT for your content and style.
Establish a pipeline to translate with DeepL using your terms, then refine with GPT using past MT edits. Enhance DeepL’s capabilities with a large language model for more precise translations.

MT engines often mishandle custom tags and entities. Custom.MT engineers can craft a tag-handling pipeline, ensuring protection during automated translations with your chosen model.
Find and quickly integrate machine translation models for the languages of India, regional variants of Arabic and Spanish, and even connect open-source models such as Meta NLLB200 and Helsinki NLP.

For small translation teams and LSP
The usage of machine translation APIs and GenAI tokens is charged on top of the platform fees. Pricing.
Teams that localize over 10 million target words per year
The usage of machine translation APIs and GenAI tokens is charged on top of the platform fees. Pricing.
Large businesses with compliance requirements
The usage of machine translation APIs and GenAI tokens is charged on top of the platform fees. Pricing.
Frequently Asked Questions
The Custom.MT MemoQ connector is a direct link that brings machine translation into MemoQ without much fuss. It operates by tying your Custom.MT account to MemoQ using an API key, giving you quick access to various MT engines. Start by creating the key in your Custom.MT Console and plugging it into MemoQ. There’s a tutorial video on connecting the Custom.MT platform to MemoQ if you want a step-by-step look.
Yes. You connect your custom or fine-tuned MT engine to the Custom.MT platform. Then you create a template using this engine and select that template when translating content from memoQ.
All setup instructions and documentation for using Custom.MT with memoQ are available here.
There’s a lineup of more than 30 MT engines and large language models available through Custom.MT, and we keep it current. You handle it all from your Custom.MT account, no extra sign-ups needed.
Yes. Set up different AI MT engines in Custom.MT, make templates for specific situations, and change them out by choosing the right one in MemoQ. This flexibility comes in handy for varied work.
Yes, you can link engines in Custom.MT templates, say one for the main translation, another for terms, and maybe one more for fixes. Link the template to MemoQ to get those layered outcomes. Give it a go in your Custom.MT account to test the difference.
Yes. You can configure templates with the required language or regional variants and use those templates when translating in memoQ. The language in the template that you want to use will need to correspond exactly to the variant you set in your memoQ project (e.g. If your memoQ project is set for German (Germany), but you have a generic German template, the system will not be able to match them.
Absolutely. You can enforce terms via Custom.MT terminology feature, by uploading your glossary directly into your templates. The machine translation output will be adapted to your terminology before the translation appears in memoQ.
Yes, but it might require some extra settings. If it doesn’t work out of the box, please contact “support@custom.mt” support@custom.mt for help.
No. Translation data is only processed and is not stored after translation. Information on Custom.MT’s data policy can be found in our Privacy policy. If third-party MT providers are used, data handling follows their respective DPAs and privacy terms.
