Gemini Embedding 2

Google's first natively multimodal embedding model

Visit WebsiteShare

Gemini Embedding 2 represents Google's pioneering natively multimodal embedding model, meticulously designed to map diverse data types including text, images, video, audio, and documents into a unified, shared embedding space. This innovative approach enables advanced multimodal retrieval and sophisticated classification capabilities across a wide array of media formats. Gemini Embedding 2 is now readily available in public preview, offering developers and researchers an unparalleled tool for building next-generation AI applications.