They say this is the most powerful AI model yet. Judging by the description, it is indeed very powerful.
The Google CEO introduction video has also been translated, and the following model is a specific introduction:
◆Gemini is multimodal, meaning it can understand, manipulate, and combine different types of information, including text, code, audio, images, and video.
◆ It is also highly flexible and can operate efficiently in a variety of environments, from data centers to mobile devices. The first version of Gemini, Gemini 1.0, is optimized for three different sizes: Gemini Ultra for highly complex tasks, Gemini Pro for a wide range of tasks, and Gemini Nano for on-device tasks.
◆Gemini Ultra surpasses current latest results on 30 of 32 widely used academic benchmarks used in the research and development of large language models. It is the first model to surpass human experts on MMLU (Massive Multitask Language Understanding), which tests world knowledge and problem-solving skills in 57 subjects such as mathematics, physics, history, law, medicine and ethics.
◆Gemini 1.0 is trained to recognize and understand text, images, audio, and more simultaneously, making it excellent at interpreting reasoning in complex subjects like math and physics. It can also understand, interpret, and generate high-quality code from popular programming languages such as Python, Java, C++, and Go.
◆Google uses its AI-optimized infrastructure and its in-house designed Tensor Processing Units (TPUs) v4 and v5e to train Gemini 1.0. The company also announced its most powerful, efficient, and scalable TPU system to date, Cloud TPU v5p, specifically designed for training cutting-edge AI models.
◆Gemini 1.0 is now gradually being applied to various products and platforms. It will be used in Google’s products like Bard and Pixel, and developers and enterprise customers can access Gemini Pro through the Gemini API in Google AI Studio or Google Cloud Vertex AI from December 13. Android developers will also be able to develop with Gemini Nano on Android 14 through AICore, a feature that will be available starting with Pixel 8 Pro devices.