Gemini
Google's family of AI models, and the assistant app that carries the same name.
Gemini is the name Google uses for two related things: a family of large language models built by Google DeepMind, its AI lab, and the chat assistant people use to talk to them. Google announced the models in December 2023. A few months later, in February 2024, it renamed its earlier chatbot, Bard, to Gemini, so the app and the models now share one name.
The models were designed from the start to be multimodal: the same model can take in text, photos, audio, video and code, rather than bolting separate systems together. They come in sizes for different jobs, with names like Pro for the most capable, Flash for fast and cheap, and Nano, small enough to run on a phone. Google also publishes Gemma, a smaller sibling whose weights anyone can download.
What makes Gemini distinct is where it lives. Besides the Gemini app on the web and phones, it is built into Google's own products: Android phones, Gmail, Docs and the rest of Google Workspace, Chrome and Search. Developers reach it through the Gemini API and through Google Cloud. Like any language model it can be wrong while sounding confident, so answers that matter deserve a second look.
Keep in mind
- Made by Google DeepMind; the models were announced in December 2023.
- The Bard chatbot was renamed Gemini in February 2024.
- Multimodal by design: text, images, audio and video in one model.
- Built into Android and Google Workspace, and available as an API.
