Bring multimodal semantic search to the edge with EmbeddingGemma 2- Google Developers Blog

Introduces EmbeddingGemma 2, a multimodal embedding model for on-device applications, supporting text, image, video, and audio. Highlights its use in Google AI Edge Gallery for instant media search and Video Moments Finder, and in Google AI Edge Foresight for context-aware meeting assistance. Details integration with ML Kit for Android and MediaPipe Tasks (Universal Embedder, Semantic Retriever, Decision Task) for cross-platform development. Mentions LiteRT as the underlying engine for unified acceleration. Provides performance metrics for visual embeddings on a MacBook M5 Pro GPU.