Introducing EmbeddingGemma: The Best-in-Class Open Model for On-Device Embeddings

EmbeddingGemma, a new 308 million parameter open embedding model, delivers best-in-class performance for on-device AI applications. This compact model ranks highest among open multilingual text embedding models under 500M parameters, supports 100+ languages, runs offline on everyday devices using less than 200MB of RAM, and enables privacy-focused applications like mobile RAG pipelines and semantic search without requiring internet connectivity.