Google’s ambitious language model, the tech giant is already making waves with its latest iteration, Gemini 1.5. Positioned as a versatile business tool and personal assistant, Gemini 1.5 Pro boasts a remarkable 87% performance boost over its predecessor, Gemini 1.0 Pro, in tests.
Utilizing the innovative “Mixture of Experts” (MoE) technique, the model processes specific requests with a segmented approach, enhancing speed and efficiency for users. The most striking feature of Gemini 1.5 is its colossal contextual window, capable of handling a staggering 1 million tokens, allowing users to pose queries based on extensive information, equivalent to 10-11 hours of video or tens of thousands of lines of code.
Google CEO Sundar Pichai emphasizes the significance of the vast context window, particularly for businesses. Filmmakers can seek reviews for entire movies, and companies can efficiently analyze extensive financial records, marking a significant breakthrough in AI capabilities.
For now, Gemini 1.5 is exclusively available to business users and developers through Google’s Vertex AI and AI Studio, eventually replacing Gemini 1.0. The standard version, Gemini Pro 1.5, with a contextual window of 128,000 tokens, will be accessible to the general public, albeit at an extra cost for the million-token version. Security and ethical boundaries are under thorough examination, especially concerning the enlarged contextual window.
Amidst the race to dominate the AI industry, Google faces competition from OpenAI, which recently announced memory for ChatGPT and is set to introduce its own web search. While Gemini shines within the Google ecosystem, challenges persist on multiple fronts.
In a strategic move, Google rebrands its AI-based chatbot Bard as Gemini, aligning with the model it’s built upon. Currently running on the Pro 1.0 model across 230 countries, Gemini introduces two enhancements – Gemini Advanced and a mobile app.
Gemini Advanced taps into the capabilities of Google Ultra 1.0’s advanced AI, excelling in complex tasks such as coding and logical thinking. With a focus on longer, more detailed conversations and improved contextual understanding, Gemini Advanced serves as a personal tutor, aiding in advanced coding scenarios and content creation.
Available in over 150 countries and part of the new Google One AI premium plan, Gemini Advanced introduces enhanced multimodal capabilities, interactive coding features, and deeper data analysis. Google is also rolling out mobile experiences for Gemini and Gemini Advanced, providing on-the-go assistance for tasks like image-based queries and content creation.
The Google One AI premium plan, including Gemini Advanced, is available for a free trial for the first two months, priced at $19.99 per month thereafter. Access to Gemini on iOS will be integrated into the Google app in the coming weeks, with Android and iOS availability expanding gradually, starting in the US and later in multiple languages and regions.
In the rapidly evolving landscape of artificial intelligence, Google’s Gemini 1.5 and Gemini Advanced mark significant strides, setting the stage for a future where users seamlessly engage with AI experiences without dwelling on the underlying technology.
Source: The Verge https://www.theverge.com/2024/2/15/24073457/google-gemini-1-5-ai-model-llm
