A German-English bilingual specialist that converts text into dense vector representations, optimized for semantic similarity tasks in both languages. It handles longer context windows than many embedding models, making it comfortable with paragraph-length inputs rather than just short phrases. Performance drops noticeably outside its two target languages.