In a groundbreaking move, Google has recently introduced Gemini, its latest and most advanced artificial intelligence model to date.

Gemini 1.0 comes in three different sizes: Gemini Ultra, Gemini Pro, and Gemini Nano, catering to diverse needs ranging from data centers to mobile devices. Among these, Gemini Ultra stands out as the largest and most powerful model.

Built with a multi-modal approach, Gemini can comprehend and integrate various types of information, including text, code, audio, images, and video. According to Google’s test results, the most powerful version, Gemini Ultra, scored an impressive 90% in the Massive Multitask Language Understanding (MMLU) test. This model synthesizes knowledge from 57 different disciplines, surpassing human expertise, which scored 89.8% in a similar test. In comparison, GPT-4 scored 87%, LLAMA-2 achieved 68%, and Anthropic’s Claude 2 reached 78.5%.

Furthermore, Gemini Ultra scored 59.4% in the Massive Multimodal Understanding (MMMU) test, surpassing 30 out of 32 benchmarks in language model research and development.

Demis Hassabis, CEO of Google DeepMind and representative of the Gemini team, stated that the company’s goal is to build a new generation of AI models that are more useful and intuitive, acting as collaborators for users.

Apart from its robust performance, Gemini 1.0 is trained to recognize text, images, audio, and various data simultaneously, enhancing its understanding of complex information and providing answers to intricate questions. The model is also capable of explaining and writing code in languages such as Python, Java, C++, and Golang.

Gemini Ultra, designed for the most complex tasks, is currently undergoing testing, while Gemini Nano is tailored for mobile device applications. The Pixel 8 Pro is the first device equipped with this AI and will offer additional features, such as summarizing recorded content and providing intelligent responses on the Gboard keyboard. Google plans to release these two versions to the market next year.

Meanwhile, the Pro version is already in use in the Bard chatbot, allowing users to interact through requests related to reading comprehension, summarization, reasoning, programming, and planning. While Bard, powered by Gemini Pro, is available in 180 countries and territories, it currently supports only the English language.