Google and Alphabet CEO Sundar Pichai has unveiled Gemini, a new artificial intelligence (AI) model that is being hailed as Google's most advanced yet. The model is a result of large-scale collaborative efforts by teams across Google, including Google Research, and is set to redefine the AI landscape with its state-of-the-art performance across a multitude of leading benchmarks.

Gemini 1.0 comes in three different sizes: Ultra, Pro, and Nano, each optimised for varying use cases. The model is multimodal, meaning it can understand, operate across, and combine different types of information including text, code, audio, image, and video. This capability allows Gemini to understand, explain, and generate high-quality code in the world's most popular programming languages, such as Python, Java, C++, and Go.

The performance of Gemini Ultra, the largest of the three, exceeds current state-of-the-art results on 30 of the 32 widely-used academic benchmarks used in large language model (LLM) research and development. Notably, Gemini Ultra is the first model to outperform human experts on MMLU (massive multitask language understanding), a significant milestone in AI development.

In addition to its language understanding capabilities, Gemini Ultra also achieves a state-of-the-art score of 59.4% on the new MMMU benchmark, which consists of multimodal tasks spanning different domains requiring deliberate reasoning.

Gemini was trained on Google's in-house designed Tensor Processing Units (TPUs) v4 and v5e. Alongside the launch of Gemini, Google is also announcing the most powerful, efficient, and scalable TPU system to date, Cloud TPU v5p, designed specifically for training cutting-edge AI models.

Google has also emphasised the comprehensive safety evaluations that Gemini has undergone, including for bias and toxicity, claiming it to be the most comprehensive of any Google AI model to date.

Gemini 1.0 is now rolling out across a range of products and platforms. Starting on December 13, developers and enterprise customers can access Gemini Pro via the Gemini API in Google AI Studio or Google Cloud Vertex AI. Gemini Ultra will be made available to select customers, developers, partners, and safety and responsibility experts for early experimentation and feedback before rolling it out to developers and enterprise customers early next year.