Google Debuts EmbeddingGemma 2: A 740M-Parameter Multimodal Embedding Model for On-Device AI

Google DeepMind has released EmbeddingGemma 2, an open-source embedding model that maps text, code, images, video, and audio into a single unified vector space—designed for privacy-preserving, on-device applications. The 740M-parameter model with modular encoder architecture runs efficiently on phones, laptops, and edge devices, supporting an 8K-token context window and achieving a 14% improvement in code retrieval tasks compared to the original EmbeddingGemma released in September 2025.