Shortly after the launch of Gemini, Google’s ambitious language model, the tech giant is already taking a leap forward with the introduction of Gemini 1.5. This upgraded version is being positioned as a versatile business tool, a personal assistant, and everything in between. The successor to Gemini 1.0 Pro, Gemini 1.5 Pro, boasts an impressive 87% improvement over its predecessor and rivals the performance of Gemini Ultra.
Utilizing the “Mixture of Experts” (MoE) technique, Gemini 1.5 runs only the necessary portions of its model when processing requests, ensuring faster and more efficient usage. The standout feature is its colossal contextual window, accommodating a staggering 1 million tokens—enabling the model to handle extensive queries and process vast amounts of information simultaneously.
Google CEO Sundar Pichai emphasizes the business applications of Gemini 1.5, citing its ability to add personal context and information at the time of query. Filmmakers can inquire about reviews for an entire movie, while companies can sift through massive financial records concurrently. Pichai hails this as one of Google’s significant breakthroughs.
For now, Gemini 1.5 is exclusive to business users and developers through Vertex AI and AI Studio. The standard version, Gemini Pro 1.5, available to the general public, comes with a contextual window of 128,000 tokens, but users can opt for the million-token version at an additional cost. Google is rigorously testing the model’s security and ethical boundaries, especially regarding the expanded contextual window.
In the race to dominate the AI landscape, Google faces competition from the likes of OpenAI, which recently announced memory for ChatGPT and is set to introduce its own web search. While Gemini appears promising, there’s still considerable work ahead for all players.
Last year’s introduction of the AI-based chatbot Bard is also part of this evolving narrative. Google has decided to rename Bard to Gemini, aligning it with the underlying AI model. Gemini, currently running on the Pro 1.0 model in over 230 countries and territories, now introduces two features: Gemini Advanced and a mobile app.
Gemini Advanced taps into Google Ultra 1.0’s advanced AI model, excelling in complex tasks such as coding, logical thinking, and collaborative projects. Serving as a personal tutor, it tailors instructions and discussions to the user’s learning style, making it invaluable for advanced coding scenarios and content creation.
Available in over 150 countries and initially in English, Gemini Advanced is part of the new Google One AI premium pricing plan. Subscribers will soon enjoy Gemini integration in Gmail, Docs, Presentations, Tables, and more.
Expanding accessibility, Google rolls out mobile experiences for Gemini and Gemini Advanced via new Android and iOS apps. Users can interact through typing, talking, or adding images, providing on-the-go assistance for various tasks. The Gemini app, available for free trial with Google One AI, offers a blend of Google Assistant features, including setting timers, making calls, and controlling smart home devices.
Gemini is set to appear on Android and iOS phones in the US initially, with a global rollout in the coming weeks. The $19.99 per month Google One AI subscription on iOS promises seamless integration with Gemini, marking a significant step forward in Google’s AI ecosystem.
Source: The Verge
