In a landmark announcement, Google has revealed "Gemini," its groundbreaking and highly versatile AI model set to redefine the frontiers of artificial intelligence. This latest innovation is poised to revolutionize AI applications, offering unparalleled capabilities across diverse platforms, from robust data centers to portable mobile devices.
Gemini encompasses three distinct versions - Ultra, Pro, and Nano - each tailored for specific tasks, ranging from complex computations to on-device efficiency. Google's rigorous testing has demonstrated Gemini's supremacy, showcasing remarkable performance that surpasses prevailing benchmarks in a multitude of domains.
- Gemini Ultra — our largest and most capable model for highly complex tasks.
- Gemini Pro — our best model for scaling across a wide range of tasks.
- Gemini Nano — our most efficient model for on-device tasks.
Notably, Gemini Ultra has secured an unprecedented 90.0% score on the Massive Multitask Language Understanding (MMLU), surpassing human experts in its grasp of 57 subjects encompassing diverse fields like mathematics, physics, history, law, medicine, and ethics. Its mastery extends to multimodal tasks and image recognition, displaying an innate proficiency without reliance on object character recognition systems.
Google's Gemini breaks barriers, diverging from conventional approaches to multimodal models by seamlessly integrating different modalities right from its inception. This unique design empowers Gemini to comprehend and reason across varied data types, marking a monumental leap in AI reasoning capabilities compared to existing models.
Nearly eight years into our journey as an AI-first company, the pace of progress is only accelerating: Millions of people are now using generative AI across our products to do things they couldn’t even a year ago, from finding answers to more complex questions to using new tools to collaborate and create. At the same time, developers are using our models and infrastructure to build new generative AI applications, and startups and enterprises around the world are growing with our AI tools. Stated by Google and Alphabet CEO Sundar Pichai.
Gemini's multifaceted abilities extend to sophisticated reasoning, simultaneous processing of text, images, and audio, and proficiency in generating high-quality code across multiple programming languages. Its prowess in competitive programming problem-solving is unparalleled, demonstrating substantial improvements over previous AI models.
Crucially, Google has placed paramount emphasis on safety and responsibility in Gemini's development. The model undergoes comprehensive safety evaluations, encompassing bias and toxicity considerations. Google is collaborating with external experts to stress-test Gemini across various parameters and has built dedicated safety measures to ensure inclusivity and content safety.
The integration of Gemini across Google's products heralds a new era of AI integration. Bard, Google's AI-powered language model, will harness Gemini Pro's enhanced capabilities, marking a significant upgrade. Additionally, Gemini Nano will feature in the Pixel 8 Pro, facilitating new features like Summarize in the Recorder app and Smart Reply in Gboard. It will be available in English in more than 170 countries and territories, and we plan to expand to different modalities and support new languages and locations in the near future.
Google aims to make Gemini available to developers and enterprise users, offering Gemini Pro via Google AI Studio and Google Cloud Vertex AI starting on December 13. Android developers will access Gemini Nano via AICore in Android 14 on Pixel 8 Pro devices.
While Gemini Ultra undergoes extensive checks, Google anticipates a broader release early next year. The company envisions Gemini as a transformative force in AI innovation, continually evolving its capabilities and applications across diverse domains, promising to reshape global lifestyles and work processes.
The advent of Gemini signifies a monumental leap forward in AI capabilities, offering unprecedented potential for innovation and transformation across industries, reaffirming Google's commitment to pioneering responsible AI advancements.