EmbeddingGemma 2 launches as an open multimodal model for on-device search

Google has launched EmbeddingGemma 2, a 740-million-parameter model released under an Apache 2.0 license. Google says it can create embeddings—representations used for search and retrieval—from text, code, images, video and audio on a device, including offline. Model weights are available on Hugging Face and Kaggle.