By Consultants Review Team
Google introduced its second-generation Gemini AI model, claiming that it will pave the way for future AI agents. The Gemini 2.0 model improves multimodality with native image and audio output, adds tools, and expands functionality. Google CEO Sundar Pichai described the new paradigm as follows: "If Gemini 1.0 was about organising and understanding information, Gemini 2.0 is about making it much more useful."
In addition to the new paradigm, Google added a function called Deep Research. This mode serves as a research helper, studying complex topics and preparing reports on the user's behalf.
The Gemini 2.0 Flash is the first publically available model in the Gemini 2.0 series. Google billed it as the series' workhorse, with low latency and improved performance. The model outperforms the flagship Gemini 1.5 Pro in key benchmarks.
It accepts multimodal inputs such as photos, video, and audio, and produces multimodal outputs such as natively generated images, mixed text, and multilingual text-to-speech audio. Gemini 2.0 Flash can also leverage Google Search, user-defined functions, and run code.
Gemini 2.0 Flash is an experimental model that developers can use through the Gemini API in Google AI Studio and Vertex AI.
AI agents with Gemini 2.0 Flash
The Gemini 2.0 Flash paradigm provides multimodal reasoning, long-context knowledge, and native tools to support agentic experiences. Google has unveiled new prototypes for AI agents, including:
An update to Project Astra, as shown during the Google I/O conference.
Project Mariner investigates future human-agent interactions.
Jules is an AI-powered coding assistant for developers.
Gemini 2.0 in AI Overviews
AI Overviews in Google Search, which delivers AI-generated summaries of search subjects, now includes Gemini 2.0. This integration improves reasoning capabilities by enabling it to handle difficult themes, advanced mathematics, multimodal queries, and coding.
Deep Research
Deep Research is a new AI mode that operates on the Gemini 1.5 model. It serves as a research helper, looking into complex topics on behalf of users.
When Deep Research receives a prompt, it develops a multi-step research plan that requires user approval. Users can modify the strategy, following which the AI examines data from the internet. It refines its findings through iterative searches before producing a detailed report with connections to original sources.
The feature is being rolled out to Gemini Advanced members on the web version and will be available on the Gemini mobile app by early 2025.
We use cookies to ensure you get the best experience on our website. Read more...
