Quality and capabilities vary between machine translation engines and over time. Through rigorous testing with multiple MT providers, you can establish which engine is best for your content. However, setting up an account with each MT provider can be cost-ineffective and cumbersome. The testing process is lengthy, labor-intensive, and expensive. Fortunately, Smartling offers functionality to determine which engine is best suited to translate your content.
Smartling Auto Select is Smartling's proprietary MT profile. It evaluates your content based on multiple criteria—including content type, target languages, linguistic and non-linguistic elements, and integrations—and routes the content to the best-suited MT engine for machine translation. Content can be translated by any of the supported MT or LLM providers, saving you time and effort in setting up and testing each provider individually.
If Smartling Auto Select's chosen engine cannot provide a quality translation, it automatically retries using the next best-suited engine. If translation is still not possible after two attempts, the content is routed to the Fallback Method, typically a human translation workflow step.
Smartling Auto Select's evaluation continuously evolves as MT engines improve, ensuring your content is always routed to the best-in-class solution for each evaluation.
For all languages, Auto Select chooses the best-performing general engine that supports the target language.
As of Q1 2026, our research shows that LLMs with a targeted prompt, especially with RAG, outperform Auto Select for the vast majority of languages. Continue below for more details.
Smartling Auto Select vs. LLMs
Our latest research shows that LLMs with a targeted prompt, and ideally RAG (which provides examples drawn from TMs, glossaries, and style guides), outperform Auto Select for the vast majority of languages. LLMs are also more controllable and flexible, since prompts can be modified at any time and will continue to improve dynamically as TMs and glossaries grow. For these reasons, we recommend using LLMs (with or without RAG) instead of Auto Select.
Learn about using LLMs in Translating with LLMs in Smartling.
Which translation providers does Smartling Auto Select use?
Smartling Auto Select uses the following translation providers:
- Amazon Translate
- DeepL
- Google Translate
- Microsoft Translator
- Prompsit (for adaptation)
Ready to translate with Smartling Auto Select?
Smartling Auto Select is available as an MT Profile in the AI Hub. No credentials are required to use this MT Profile.
Smartling Auto Select works like any other MT Profile. You can use it in workflows, set it as the MT Profile for MT suggestions in the CAT Tool, and use it to translate content through Smartling's MT API or through other instant MT integrations.