A compact, focused specialist built for generating text embeddings rather than producing conversational output. With a 514-token context window, it operates within tight constraints, making it suited for shorter text passages rather than long documents. Its open-weight nature means the internals are fully accessible for inspection or fine-tuning.