A compact multilingual workhorse that maps text from 15 languages into a shared embedding space, making cross-lingual similarity comparisons straightforward. It trades some representational depth for speed and size, running efficiently on modest hardware. Particularly comfortable with sentence-level semantic tasks rather than long document understanding.