# Search across images and text with Gemini file search

> Agno now supports multimodal inputs in the Gemini File Search API, so agents can index and semantically search across images alongside text rather than text alone.

- Published: 2026-05-06
- Author: Agno Team
- Category: Changelog
- Canonical: https://agno-com-nine.vercel.app/articles/search-across-images-and-text-with-gemini-file-search
- Markdown: https://agno-com-nine.vercel.app/articles/search-across-images-and-text-with-gemini-file-search.md

Agno now supports multimodal inputs in the Gemini File Search API, so agents can index and semantically search across images alongside text rather than text alone. This is available with `google-genai≥1.75.0`; older versions remain text-only.

**Action required:** Bump `google-genai` to `1.75.0` or later to use image inputs with Gemini file search. Existing text-only setups keep working without changes.

See the [cookbook example](https://github.com/agno-agi/agno/blob/main/cookbook/90_models/google/gemini/file_search_image_upload.py) for an image-upload walkthrough.