Jump to Content

Our largest model for highly complex tasks

1.0 Ultra is optimized for high-quality output across complex tasks like code and reasoning, with support for multiple languages.

1.0 Ultra is optimized for high-quality output across complex tasks like code and reasoning, with support for multiple languages.

Ultra capable

State-of-the-art performance across a range of industry benchmarks.

Multimodal reasoning

Natively understands and reasons across sequences of audio, images, and text.

Complex coding

Excels at coding and achieves state-of-the-art performance when integrated into AlphaCode 2.

Mathematical reasoning

Advanced analytical capabilities and strong performance on competition-grade problem sets.

The first model to outperform human experts on MMLU (Massive Multitask Language Understanding), a benchmark that uses 57 subjects including math, physics, history, law, medicine and ethics to test both world knowledge and problem-solving ability.

Demo

Ultra in action

Learn what our Gemini models can do from some of the people who built them.

Research

Technical reports

May 2024 Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context
December 2023 Gemini 1.0: A family of highly capable multimodal models

For developers

Build with Gemini

Integrate Gemini models into your applications with Google AI Studio and Google Cloud Vertex AI.

Get the latest updates

Sign up for news on the latest innovations from Google DeepMind.

Explore our other teams and product areas