Google has added multimodal search to Gemini API's File Search.

It can now search images and text together, also supports custom metadata filtering, and provides page-level citations.
People doing RAG should be able to use it immediately.
What I value most is that it finally starts handling hybrid data scenarios; putting visual materials, contract versions, and knowledge base statuses into the same search process will be much more efficient.
Those working on customer service knowledge bases, legal document retrieval, or content asset libraries can take a look first.
The official documentation is here: