Google DeepMind releases 740M EmbeddingGemma 2 under Apache 2.0 license

Google DeepMind has released EmbeddingGemma 2 under an Apache 2.0 open-source license. The model has 740 million parameters and is designed for on-device multimodal embeddings, with model weights hosted on Hugging Face and Kaggle. According to Google DeepMind, the 740M model is competitive across benchmarks and outperforms some specialist models more than twice its size.

The model enables developers to implement multimodal search, such as matching video segments via voice memos, or combine it with Gemma 4 for private on-device retrieval-augmented generation (RAG). Additional technical details and links are available in Google DeepMind's [overview documentation](https://goo.gle/4y9wXgl).