Gemini

Gemini is Google DeepMind's family of multimodal models — current generations split into Flash (fast, high-volume), Pro (complex reasoning) and Deep Think (research-grade) tiers — trained to read and generate text, images, audio and video from a single prompt, and increasingly built for multi-step agentic work rather than single-turn chat (deepmind.google/models/gemini). It reaches users three ways that matter for tracking: the consumer Gemini app, the Gemini API that developers build directly on, and as the model Google plugs into Search's AI Overviews and AI Mode. That last point is where it is easiest to conflate two different things. "AI Overviews" names a Search surface, not a model — Gemini is what typically powers it, but a citation inside an Overview and a brand mention inside a standalone Gemini chat response are separate events. A visibility tool that only checks one is not measuring the other, which is why tools in this corpus such as Ahrefs Brand Radar and Archytas AISpy list Gemini as a tracked surface distinct from "Google AI Overviews."