Google Introduces 'Gemini', a Powerful Generative AI Model

Google, on Wednesday, December 6, 2023, debuted its most capable and very own generative AI model, Gemini.

Demis Hassabis, the CEO and co-founder of Google DeepMind, claims that teams from all over Google, including their colleagues at Google Research, worked extensively to create Gemini. Google unveils the potent Gemini generative AI model. It was designed from the ground up to be multimodal, which means it can comprehend, function across, and blend various forms of input, including text, code, audio, image, and video.

He also added that Gemini is also their most flexible model yet, able to efficiently run on everything from data centers to mobile devices. Its state-of-the-art capabilities will significantly enhance the way developers and enterprise customers build and scale with AI.

​Gemini 1.0, Google's first version, is optimized for three different sizes:

​Gemini Ultra

Google's largest and most capable model for highly complex tasks.

Gemini Pro

Google's best model for scaling across a wide range of tasks.

Gemini Nano

Google's most efficient model for on-device tasks.

Google claims to have been thoroughly testing and assessing the performance of their Gemini models on a wide range of tasks. Gemini Ultra outperforms the state-of-the-art findings on 30 of the 32 commonly used academic benchmarks used in large language model (LLM) research and development, from natural image, audio, and video understanding to mathematical reasoning.

In MMLU (Massive Multitask Language Understanding), a widely used technique to assess the expertise and problem-solving skills of AI models, Gemini is the first model to surpass human experts with a score of 90.0%.
​
Also, Google claims that by utilizing their new benchmark approach for MMLU, Gemini can make considerable gains over merely relying on its initial impression by using its reasoning capacities to deliberate more thoroughly before responding to challenging issues.

AI has the potential to be one of the most important
and beneficial technologies ever invented.
DEMIS HASSABIS, CO-FOUNDER AND CEO, GOOGLE DEEPMIND

Google DeepMind unites Google Brain and DeepMind, two of the top AI labs in the world, into a single, focused team under the direction of its CEO, Demis Hassabis. The two teams have produced some of the most significant advances in AI research in the past ten years, many of which serve as the foundation for the thriving field of artificial intelligence as it exists today.
​
Using an interdisciplinary approach to the development of general AI systems, DeepMind was founded in 2010. The research lab combined innovative approaches to structuring scientific pursuits with cutting-edge concepts and developments in machine learning, neuroscience, engineering, mathematics, simulation, and computing infrastructure.