New AI Model Identifies 61 African Languages Accurately

GSMA and French AI firm Pleias have released CommonLingua, an open-source language identification model covering 334 languages, including 61 African languages across eight language families. The two-million-parameter model achieves 83% accuracy, significantly outperforming existing tools that routinely mislabel African-language text as English or French. The model processes UTF-8 byte sequences directly, supporting scripts including Arabic, Ethiopic and N'Ko. Trained on open-licensed content, it aims to lay foundational AI infrastructure for a continent home to between 2,000 and 3,000 languages, where existing AI systems lose roughly 30 percentage points in accuracy compared to major world languages.