Google

Gemini

Multimodal reasoning tuned for research and grounded answers.

Build a Gemini prompt

Overview

Gemini is Google's multimodal model family with especially strong image, video, and code understanding. It's a natural fit when you want the model to reason over screenshots, diagrams, or spreadsheets.

Best prompts for Gemini

  • Data extraction from images
  • Research with citations
  • Video understanding
  • Sheet formulas

Tips

  1. 01

    Attach the source instead of describing it — Gemini reads images natively.

  2. 02

    Ask for citations when factuality matters.

FAQs

Can Gemini browse the web?
Yes, when grounded-search is on. Ask it to include sources so you can verify.