Google’s Gemini 2.0 has new features and capabilities. These include improved multimodal understanding, agentic AI, increased speed, better battery life (even for phones with excellent batteries), and broader integration with other Google solutions. Gemini 2.0 processes information differently than its predecessor and achieves more complex tasks.
Integrations with Google products such as Search, Maps, and Workspace are key focus areas, although some features are still rolling out. Gemini 2.0 is accompanied by a major UI update to NotebookLM, Google’s Gemini-powered AI information warehouse that leverages your research materials, links, and datasets.
Related
Google Gemini: Everything you need to know about Google’s next-gen multimodal AI
Google Gemini is here, with a whole new approach to multimodal AI
5 Native image and audio processing
Eliminating translation promises better responses
Unlike previous models, which required converting images and audio into text before analysis, Gemini 2.0 processes them. The goal is to eliminate the information loss associated with translation. Direct processing allows a richer, more nuanced understanding of the input, capturing subtleties and contextual cues that would otherwise be lost. Gemini 2.0 promises a more accurate and efficient interpretation of multimedia content by bypassing the intermediary text conversion step.
Gemini 2.0 identifies objects in an image and understands their relationships and the scene context. I tested its abilities, and the response was detailed and accurate. It even recognized the materials from which objects on my coffee table were constructed. I also ran the image through version 1.5 Pro. While it provided some of the same information, its response was less detailed. The Gemini 2.0 Flash model still refused to process an image with people.
If Gemini 1.0 was about organizing and understanding information, Gemini 2.0 is about making it much more useful. – Sundar Pichai, Google CEO
4 Agentic AI
Gemini 2.0 can do more with less
Agentic AI describes AI models that actively interact with the world to achieve specific goals. Gemini 2.0 powers AI agents, allowing them to execute complex, multistep tasks that require planning, decision-making, and interaction with external systems. Agentic AI may mark a turning point where AI becomes a more proactive problem-solver.
Gemini 2.0’s agentic capabilities are slated to integrate with external tools like Google Search, Maps, and Lens. For example, a Gemini 2.0 AI agent could leverage Google Maps to plan a complex itinerary involving multiple destinations and modes of transportation. However, this functionality wasn’t available to me in the 2.0 Flash desktop or from Maps. Google recently rolled out 2.0 in a pre-release version of its mobile app, which is where we expect to see some of these capabilities shine.
In its blog post, Google discusses how the new model relates to two major Google initiatives: Project Astra and Project Mariner. Project Astra focuses on agentic AI capabilities integrated with services such as Search and Maps. Project Mariner touches on automated web features such as filling out forms, booking reservations, and gathering information from multiple websites.
3 Deeper integrations across the Google ecosystem
AI goes everywhere with Gemini 2.0
Gemini 2.0 integrates deeply across Google’s ecosystem of products and services. The promise is a more unified and seamless user experience. Gemini 2.0’s extended integrations point toward Google’s strategy of using Gemini as a common thread woven throughout Workspace.
Google Search is getting deeper integration with Gemini 2.0, facilitating more conversational search experiences and leveraging AI Overviews for comprehensive answers to complex queries, as we predicted in early November. Within Google Workspace, AI-powered features driven by Gemini 2.0 are being incorporated into applications like Docs, Slides, and Meet to enhance productivity and collaboration. Android Assistant is set to receive new capabilities powered by Gemini 2.0. Your mileage may vary during the rollout process.
2 Faster responses and better battery life
Gemini 2.0 Flash doubles the speed of 1.5
The full name of the latest version is Gemini 2.0 Flash Experimental. It’s been streamlined for speed and responsiveness. Gemini 2.0 Flash delivers enhanced performance while reducing latency. This positions Gemini 2.0 Flash to better power real-time multimodal interactions.
Gemini 2.0 Flash claims notable performance improvements. Google says it’s twice the speed of its predecessor. In my experimentation, responses were nearly instantaneous. They were markedly faster than when I fed the same queries to version 1.5 Pro. The faster response times make interactions feel natural and fluid. For audio conversations, the reduced latency could reduce delays and create a more engaging and realistic experience.
Gemini 2.0 Flash might extend the battery life for AI processes on mobile devices such as your Google Pixel 9 or other smartphone. This could mean less frequent charging, something everyone can appreciate.
1 NotebookLM’s reinvented UI
Gemini 2.0 is accompanied by a redesign of NotebookLM’s interface and new features
It isn’t in Gemini 2.0, but the two are different sides of the same coin. The arrival of Gemini 2.0 marks a parallel iteration in NotebookLM. The iteration goes beyond its underlying AI capabilities and into its user interface. The overhaul seeks to make it more intuitive and efficient for users to interact with their notes and documents. It focuses on streamlining workflows, improving navigation, and providing a more refined visual environment.
Related
I tested NotebookLM and saw the promise of something great
NotebookLM has a lot of issues, but the idea behind it has potential
Gemini moves fast and isn’t slowing down
Gemini 2.0 has cool tricks for maximum productivity. Along with recognizing text, it also understands images and sounds. This version promises to do things for you, like using Google Search or Maps to find information or complete complex tasks. Moreover, it has a larger context window than its predecessor. Google pegs Gemini 2.0 Flash at 2 million tokens, meaning it retains and processes twice as much information as Gemini 1.5 Pro.
By focusing on multimodal understanding, agentic capabilities, deeper integrations with Google apps, and performance enhancements, Google is making Gemini the foundation of its ecosystem. As mainstream AI continues to mature, 2025 will be an interesting year.
