Supported MT and LLM Providers

Smartling supports a wide range of Neural Machine Translation (NMT) providers and Large Language Models (LLMs), including a pre-configured option called Smartling Auto Select, which automatically routes your content to the best-suited MT engine for translation. 

You can use MT engines or LLMs for translation in various ways, including within translation jobs, linguist suggestions in the CAT Tool, instant file or text translation using Smartling Translate, Smartling's MT API, or through one of Smartling's MT integrations that deliver translations directly where they’re needed. To learn more, see Introduction to MT and AI Translation in Smartling. 

The use of MT or LLM translation falls under Smartling's AI Hub. Please note that this is a paid product. For pricing information, please reach out to your Smartling Customer Success Manager.

Supported Providers 

To connect your Smartling account to an MT engine or LLM, you will first need to activate an account with one of the above providers. Please contact your Smartling Customer Success Manager for more details.

Smartling also supports copy source to target, where the source text is copied over to the translation. This can be useful when performing adaptation.

Built your own in-house MT or LLM service? Use it in Smartling—learn more in Bring Your Own MT or LLM Provider.

Translating with LLMs

Large Language Models (LLMs) can be used as translation providers in Smartling, similar to Machine Translation (MT) engines. Currently, Smartling supports three LLM providers: GPT Microsoft Azure, GPT OpenAI, and Google Gemini (Vertex AI).

The process of setting up an LLM as a translation provider is the same as for an MT engine. However, when creating an MT Profile for an LLM provider, you’ll need to enter a translation prompt. The MT Profile also includes a space to test the prompt.

Unlike MT engines, which are generally ready to use out of the box, LLMs rely heavily on well-designed prompts to produce the desired translation output. Additionally, LLMs are prone to hallucinations, which refers to cases where they generate nonsensical, incorrect, or inconsistent translations. Because of this, MT engines are often a more reliable and recommended option for translation, while LLMs are better suited for smoothing and refining the translation output. 

Locale-specific machine translation

Machine translation generally facilitates the translation of one language into another without accounting for locale-specific nuances. For example, most MT providers will generate the same translation output for German (Germany) and German (Austria). Regional or country-specific variations of the same target language are generally not taken into account, though there are some exceptions, such as French (France) vs. French (Canada), Portuguese (Portugal) vs. Portuguese (Brazil), Chinese (China-Simplified) vs. Chinese (Taiwan-Traditional), and Spanish (Spain) vs. Spanish (Latin America). 

Cross-locale adaptation with Prompsit

If you are interested in adapting text from one locale to another with MT, e.g., English (United States) to English (United Kingdom), we would recommend setting up an adaptation workflow using Prompsit.

Prompsit can also be used to translate from or into the following English locales:

• English (Canada): This locale uses the same translation output as English (United States).
• English (Australia), English (Singapore), English (Mauritius), English (New Zealand) and English (South Africa): These locales use the same translation output as English (Great Britain).

Locale-specific MT with Smartling Auto Select

Smartling Auto Select provides a locale-specific machine translation output for the following target languages, taking into account regional nuances:

• French: French (France) vs. French (Canada)
• Portuguese: Portuguese (Portugal) vs. Portuguese (Brazil)
• Spanish: Spanish (Spain) vs. Spanish (Latin America)
• English: English (United States) or English (Canada) vs. English (United Kingdom) or English (Australia), English (Singapore), English (Mauritius), English (New Zealand), English (South Africa)

This is achieved by using Prompsit to automatically adapt the machine translation output to the specific target locale, regardless of the MT provider which was selected for the initial translation.

Related articles

1. Learn more about Smartling's AI Hub. 
2. Learn more about Smartling's optimized MT routing service, Smartling Auto Select. 
3. Learn about creating and managing MT profiles.