Google Unveils Gemini: Next-Gen AI Model in 3 Variants, Powers Bard & Pixel 8
Google has recently unveiled its latest language model, Gemini AI, which is available in three distinct sizes to cater to specific requirements. This artificial intelligence model is designed to comprehend and operate across diverse forms of information, such as text, code, audio, images, and video.
Gemini, introduced by Google, is touted as the company's most versatile and general-purpose AI model to date. Contrary to being a singular model, Gemini is a family of AI models, comprising three sizes: Gemini Ultra, Gemini Pro, and Gemini Nano.
According to Google, Gemini has been launched in its initial iteration, version 1.0, with each size tailored to meet specific needs and the ability to operate across a range of devices from data centres to mobile devices. Sundar Pichai, the CEO of Google, explained in an official blog post, "We designed Gemini to be natively multimodal, pre-trained from the start on different modalities. Then we fine-tuned it with additional multimodal data to further refine its effectiveness."
The three variations of Gemini are as follows:
Gemini Nano: This lightweight AI model is optimized for mobile devices, particularly the Pixel 8. It facilitates on-device tasks seamlessly, even offline, and delivers robust AI capabilities without relying on external servers. Gemini Nano is capable of suggesting relevant replies in chat apps and summarizing text content, ensuring a smooth and responsive user experience.
Gemini Pro: Positioned as the advanced variant, Gemini Pro is designed to power various Google AI services, serving as the foundational support for Bard. Running on Google's data centres, it plays a crucial role in enhancing the capabilities of the AI chatbot, Bard, offering fast response times and an exceptional ability to understand complex queries.
Gemini Ultra: Described as the most powerful LLM by Google, Gemini Ultra excels at handling highly complex tasks and is best suited for data centres and enterprise applications. Although not widely available yet, Google asserts that Gemini Ultra surpasses "current state-of-the-art results on 30 of the 32 widely-used academic benchmarks used in large language model (LLM) research and development." This positions it as a high-performing option for various standard tests, especially those involving complex tasks, and is expected to be released after completing its testing phase.
In terms of availability, Gemini AI, starting with Gemini Nano, is now accessible on Pixel 8 Pro, introducing enhanced features such as Summarise in the Recorder app and Smart Reply on Gboard, initially implemented in WhatsApp. The expansion of Gemini includes integration into various Google products and services, such as Search, Ads, Chrome, and Duet AI.
Furthermore, Gemini is already being integrated into Google Search to accelerate the Search Generative Experience (SGE), resulting in "a 40 per cent reduction in latency in English in the US, alongside improvements in quality."