A compact, openly available text model from Google's research lineup. With 400 million parameters, it operates in a lightweight footprint, making it suited for environments where resource efficiency matters. Details about its specific task strengths and behavioral tendencies are limited beyond its small-scale architecture.