Today, Google has made its latest open embedding model, EmbeddingGemma, available as open-source software. This model, which encompasses 308 million parameters, has been meticulously crafted for on-device AI applications. It can be seamlessly deployed across a range of devices, including laptops and smartphones. EmbeddingGemma is well-suited for tasks such as Retrieval-Augmented Generation (RAG) and semantic search. Among its standout attributes are its capacity to generate top-notch embedding vectors, its commitment to safeguarding user privacy, and its ability to function flawlessly even in offline mode. In terms of performance, it rivals that of the Qwen-Embedding-0.6B model, despite being only half its size.