Gemini Ultra

Gemini Ultra was announced on December 6, 2023, alongside Gemini Pro and Gemini Nano, as the first generation of Google DeepMind's Gemini model family. General availability through the Gemini Advanced subscription service followed in early 2024, after additional safety testing and red-teaming.

Gemini Ultra is accessible through Google's Gemini Advanced subscription, which is bundled with the Google One AI Premium plan. It is not available as an open-weight model for local deployment, unlike some competing large language models in the same tier.

Gemini Ultra is the largest and most capable model in the first-generation Gemini family, designed for the most complex reasoning tasks across text, images, audio, and video. Gemini Pro serves as a mid-tier model for general-purpose use, while Gemini Nano is designed to run on-device on mobile phones.

Yes, Gemini Ultra was designed from the ground up to be natively multimodal, meaning it can process and reason across text, images, audio, and video within a single model rather than relying on separate specialized models for each modality. Google DeepMind reported that this native multimodality gave it advantages over models that combined separate text and vision components.