A specialist that converts text into dense numerical representations, excelling at capturing semantic meaning across multiple languages. It operates quietly in the background — no conversation, no generation — just precise vector outputs that power search, clustering, and retrieval pipelines. At 8B parameters it sits on the heavier end for an embedding model, trading deployment simplicity for richer representational capacity.