Gemini API File Search is now multimodal: build efficient, verifiable RAG

Google has expanded the Gemini API's File Search tool with three major capabilities: multimodal support (images and text processed together via Gemini Embedding 2), custom metadata filtering for scoping queries to specific data slices, and page-level citations that tie model responses to exact source locations in documents. These updates make it easier to build production-grade RAG systems with better accuracy, transparency, and verifiability.