Google for Developers: Introducing EmbeddingGemma 2: An open model for natively multimodal embeddings

Sebastian Russo, Ivan Llanos, and Sahil Dua from Google DeepMind introduce EmbeddingGemma 2, a lightweight open model that maps text, images, video, and audio into a single unified embedding space.



At just 740 million parameters, EmbeddingGemma 2 is engineered for on-device use cases. Watch this video to discover:



*EmbeddingGemma 2’s modular architecture with optional audio and vision encoders

*Instant media search and video moment retrieval running fully offline on device

*Retrieval-augmented generation (RAG) pipelines where sensitive data never leaves the hardware

*Fine-tuning for domain-specific use cases





🔗 Resources:

Learn about the model → https://goo.gle/4hE3Lb9

Read the developer guide → https://goo.gle/4hFnyqO

Explore on-device embeddings with Google AI Edge → https://goo.gle/3VSXqS1

Download model weights from HuggingFace → https://goo.gle/4i45qbp

Get started with the Gemma Cookbook → https://goo.gle/3TCsqVO



Have questions? Drop them in the comments.







Subscribe to Google for Developers → https://goo.gle/developers







Speakers: Sebastian Russo, Ivan Llanos, Sahil Dua

Products Mentioned: Gemma, Google AI Edge