Meta Releases Open Source AI Model Capable of Translating 200 Languages
Meta launches NLLB-200, an open source AI model that translates 202 languages with significant quality improvements for historically underserved African and Indian dialects.
Meta releases an open source artificial intelligence model called NLLB-200 that breaks the 200-language barrier for machine translation. As part of the No Language Left Behind initiative, this new model aims to provide high-quality translations for languages that previous technologies largely overlook.
The NLLB-200 system shows impressive performance gains over existing state-of-the-art translation tools, achieving an average improvement of 44 percent on the BLEU standard. It delivers particularly strong results for underrepresented dialects, translating 55 African languages with high accuracy and showing up to a 70 percent improvement for certain African and Indian languages.
Meta makes the NLLB-200 model freely available on GitHub and pairs this release with $200,000 in grants for nonprofit organizations. These funds support researchers who want to explore real-world applications for the technology and help bridge the digital divide for low-resource language communities.