Model
Gemini
Google DeepMind's natively multimodal model family, introduced in December 2023 and now in its fourth generation with Gemini 4 Argon.
Gemini is the family of AI models built by Google DeepMind and the brand under which Google offers its assistant. It was introduced on 6 December 2023 by Demis Hassabis as “the most capable and general model we’ve ever built,” designed from the start to work across text, code, audio, images and video rather than bolting modalities together. The first generation shipped in three sizes, Ultra, Pro and Nano, the last intended to run on devices. Google’s earlier chatbot, Bard, was renamed Gemini as the models were adopted.
Later generations added longer context, agentic tool use and reasoning. The family is tiered as Pro, Flash and Flash-Lite, alongside dedicated image models branded Nano Banana and video and audio models under the Gemini Omni name.
The current generation is Gemini 4. Google announced it on 30 September 2026, anchored by a top-tier model named Argon, which a spokesperson called the company’s “most performant model yet built for complex workloads” and which is larger than previous Pro models. Google gave early access to cybersecurity partners and took part in a voluntary US pre-release review. Argon reached Google AI Pro and Ultra subscribers in the Gemini app on 13 October 2026. Google said it would not release the previously planned Gemini 3.5 Pro.
Gemini is accessed through the Gemini app on web, mobile and desktop; inside Gmail, Docs and other Workspace apps; in Google Search’s AI Mode; through the Gemini API via Google AI Studio; and through Google Cloud’s enterprise offerings. Google does not release Gemini weights.
For current models and pricing see gemini.google.