Meta Unveils NLLB-200 AI Model to Translate 200 Languages
Meta introduces NLLB-200, an artificial intelligence model capable of translating 200 languages with a focus on low-resource languages. The tool is already helping Wikipedia editors expand access to global information.
Meta announces NLLB-200, a new artificial intelligence model that translates content across 200 different languages. The company releases a detailed 190-page paper alongside a demo site where users test the translation tool firsthand. This milestone represents a massive step toward Meta's broader goal of eventually translating the world's 7,000 languages.
The project focuses heavily on low-resource languages, which make up roughly three-quarters of the model's 200 available languages. Meta warns that without dedicated translation efforts, the internet becomes increasingly inaccessible for speakers of these underrepresented languages. The company designs this technology to ensure real-time text translation integrates smoothly into upcoming augmented and virtual reality metaverse experiences.
NLLB-200 is already making a practical impact through a new partnership with the Wikimedia Foundation. Wikipedia editors currently use the AI tool to rapidly translate articles into low-resource languages that lack a strong presence on the site. Meta emphasizes that this technology empowers people to both consume and share information freely across language barriers.