Google Launches EmbeddingGemma 2, a 740M On-Device Multimodal Embedding Model

Google has launched EmbeddingGemma 2, an open 740 million-parameter embedding model designed to run locally while mapping text, code, images, video and audio into one shared vector space. The release moves Google's lightweight EmbeddingGemma line beyond text and gives developers a single on-device retrieval layer for private media search, code search and RAG.