Model2Vec turns any sentence transformer into a small, fast static embedding model. It produces distributional bag-of-words embeddings that maintain competitive performance while dramatically reducing model size. The technique enables efficient semantic search and retrieval without requiring full transformer inference.