The Lionbridge Machine Translation tracker is the longest-standing measure of MT in the industry.

The tracker measures the overall performance of the five major neural MT engines and several GenAI models. It also evaluates translation quality based on language pairs and domains. GenAI does not outperform the major neural MT engines, with some exceptions. However, these models produce decent results, especially considering they haven’t been trained explicitly for translations.

What’s the takeaway? Amidst the strong interest in deploying GenAI/LLMs, Machine Translation continues to prove itself to be a worthy automated translation tool.

Translation results are constantly changing, and the tracker captures these fluctuations.