DeepL said it is implementing one of Nvidia's latest systems, which will allow the German startup to translate the entire internet in just 18 days, down from 194 days previously.
DeepL said it is implementing one of Nvidia's latest systems, which will allow the German startup to translate the entire internet in just 18 days, down from 194 days previously.
DeepL, a startup that has developed its own AI models to compete with Google Translate, has become the first in Europe to use Nvidia’s system, known as the DGX SuperPOD. Each DGX SuperPOD server rack contains 36 B200 Grace Blackwell superchips, one of the company’s newest products on the market. Nvidia’s chips are essential for training and running the huge AI models, including those developed by DeepL.
New capabilities make DeepL faster and more accurate than ever before:
“The idea, of course, is to give our scientists much more computing power to build even more sophisticated models,” Stefan Mesken, chief scientist at DeepL, told CNBC.
Mesken said the updated infrastructure will help improve current products, such as Clarify, which the company launched this year. Clarify is a tool that asks users questions to ensure context is taken into account in translations.
"Not long ago, this was technically impossible, but thanks to the progress we have made in our next-generation development efforts, it has become a reality. These are the kinds of breakthroughs we continue to seek," Mesken added.
Recall that dev.ua analyzed in detail how the DeepL translator works and what its advantages are compared to Google Translate.
Comment hidden for violating commenting rules.
Comment hidden for violating commenting rules.