EmbeddingGemma 2 Expands Beyond Text to Support Code, Images, Video, and Audio

Google released EmbeddingGemma 2, a compact multimodal embedding model supporting text, code, images, video, and audio for efficient on-device semantic search and retrieval-augmented generation. It unifies embeddings across modalities in a 768-dimensional space and offers scalable parameter sizes and storage-efficient truncation.