A focused specialist that converts text into dense numerical representations, excelling at capturing semantic similarity across long passages. It handles extended context windows gracefully, making it reliable for chunking-heavy retrieval pipelines. Its strength is consistency — it doesn't generate, it compresses meaning.