Google recently launched Gemini, a new type of AI model that works with text, images, and video.
The initial version of Gemini will roll out inside of Google’s chatbot Bard for the English language setting and will be available in move than 170 countries and territories. Google described Gemini as “natively multimodal” since it was trained on images, video, and audio, instead of only text, like many other large language models are. There are also three versions; Ultra is the largest and most capable, Nano is significantly smaller and efficient, while Pro is lies right in the middle both in size and capabilities.
A technical report has also been released providing details on Gemini’s inner workings. Google is expected to have developed a novel design for the model with a new mix of training data in an effort to reestablish itself as the world’s leading AI company. There has been comprehensive quality and safety testing because of the model’s more general capabilities. To learn more about the new model, including how it was named, visit here.