Google Gemini is introducing an exciting new feature called “Deep Research,” marking a significant step toward Google’s vision of AI agents capable of performing complex tasks on your behalf. This feature, the first of its kind in Gemini, showcases advanced capabilities such as native image and audio output, multimodal functionality, and seamless tool integration.
Gemini 2.0 aims to develop highly capable AI agents, bringing Google closer to its goal of a universal assistant. First previewed at the Made by Google 2024 event in August, the feature allows users to pose research questions and receive a detailed, multi-step plan in response. Users can also customize and refine the plan by adding new aspects or areas of focus.
Gemini’s research process begins with “finding interesting pieces of information” on the internet, followed by related searches, a process it repeats multiple times. When Gemini Deep Research returns a report, it will present it to you with citations and references, separated by headline and section. Users can link directly to any source, ask follow-up questions, and refine the report as needed.
Instead of functioning as a stand-alone model, Deep Research functions as a “system of Gemini 1.5 Pro models”. When given a prompt, the first Gemini 1.5 Pro analyzes the request and devises a detailed plan. It then creates specialized versions of itself to execute the plan, processing information from hundreds of websites and public sources. However, it cannot access research papers or websites behind paywalls or the ones that require sign-ins.
Currently, Deep Research is exclusive to Gemini Advanced subscribers on the web. Go to Gemini and choose “Gemini 1.5 Pro with Deep Research” from the model dropdown menu to give it a try.
- Meta’s AI Video Editor Reimagines Short Videos Instantly - June 13, 2025
- Google Search Gets Smarter with Real-Time Data Visualisation Tools - June 9, 2025
- Bing’s New AI Video Creator Uses OpenAI’s Sora – And It’s Free - June 3, 2025
