Google has unveiled Gemini, its “largest and most capable AI model,” in three sizes: Ultra, Pro, and Nano, following its introduction at I/O 2023. Every size has a distinct purpose and can operate on a variety of platforms, including mobile phones and data centers. According to the company, the Ultra model will be released early in 2019. The Pro and Nano models are now available. The three models are briefly compared, along with Google’s intended applications, here.
Pixel 8 Pro smartphones will be powered by Gemini Nano.
Google’s most effective variant, Gemini Nano, is beginning to be available on Pixel 8 Pro smartphones and is designed for on-device operations. According to Google, the Pixel 8 Pro is the first smartphone designed for the Gemini Nano, and it will leverage the capabilities of the Google Tensor G3 chip to enable two new features: Use Gboard’s Smart Reply and Recorder’s Summarize features.
In Recorder, summarize: The recorder app is already a strong tool that lets users view the recording’s transcript instantly thanks to artificial intelligence. The Summarize function on the Pixel 8 Pro will now be powered by Gemini Nano, which the firm claims is its most efficient model designed for on-device interactions. Even in the absence of a network connection, this function of the Recorder app will give an overview of recorded talks, interviews, presentations, and more.
Gboard’s Smart Reply: As a developer preview, Smart Reply in Gboard will be powered by the Pixel 8 Pro. The on-device AI model, which is now compatible with WhatsApp and will be added to additional apps in the upcoming year, is designed to save time by providing conversationally aware suggestions for high-quality responses.
“Gemini Nano running on Pixel 8 Pro offers several advantages by design, including the ability to use features without a network connection and assisting in preventing sensitive data from leaving the phone,” the company stated.
Additionally, Google reports that additional features for the Assistant with Bard experience on Pixel will be powered by the larger family of Gemini devices early in 2019.
Google Bard’s Gemini Pro
The “best model for scaling across a wide range of tasks,” according to Google, is Gemini Pro. It will enable the Bard AI chatbot to do tasks including more complex planning, comprehension, and reasoning.
According to Google, Gemini Pro beats GPT-3.5 on six of the eight industry-standard metrics, including grade school math reasoning and Massive Multitask Language Understanding (MMLU).
Additionally, Google has improved the capabilities of Gemini Pro in Bard to include planning, coding, understanding, summarizing, and reasoning. More than 170 nations and territories, including India, have access to it in English.
Bard Advanced will receive Gemini Ultra in the upcoming year.
Beginning with Bard Advanced early in the upcoming year, Gemini Ultra—Google’s largest and most capable model for extremely complicated tasks—will be accessible. “New, cutting-edge AI experience that gives you access to our best models and capabilities” is what this is advertised as.
According to Google, thorough safety and trust tests are presently being completed for Gemini Ultra. After more development, fine-tuning, and reinforcement learning based on user feedback, the model will be released.
- Tata Group Plans to Construct one of India’s Biggest iPhone Manufacturing Plants: Location, Number of Jobs to be Created, and Other Facts
- How To Use Paytm To Make Credit Card Payments Via UPI
- What Are The New SIM Card Regulations That Take Effect on December 1st?
- On E-Commerce Websites, The Government Has Prohibited 13 “Dark Patterns”: How Do They Look?