Home
Google's Gemini 3.1 Pro Deep Research Agent Launches with MCP Protocol and Multimodal Capabilities
Google recently unveiled two new self-research agents built on the Gemini 3.1 Pro architecture: Deep Research and Deep Research Max. These tools are now accessible for public preview via the paid tier of the Gemini API. They are designed to fully automate intricate research workflows, signaling a move for AI agents from basic web queries toward "long-term reasoning" models capable of deep analysis.

The standard Deep Research agent emphasizes efficiency and low latency, making it ideal for real-time conversational applications that need quick responses. In contrast, Deep Research Max focuses on depth of investigation, utilizing extended processing time for multi-step reasoning and iteration, primarily aimed at asynchronous backend tasks like due diligence reports. Technically, this new iteration introduces support for the Model Context Protocol (MCP), enabling agents to pull data from the open web as well as private databases containing financial or market information. Furthermore, the agents now feature native visualization tools, allowing them to directly generate charts and infographics in HTML format.

In benchmark performance tests, Google reports that Deep Research Max demonstrates substantial gains in retrieval and reasoning tasks compared to previous models. However, industry analysts note that comparisons with the OpenAI GPT-5.4 series and Anthropic's Opus4.6 can be influenced by testing methodologies and should be viewed with caution. A notable addition is a collaborative planning feature, supporting multi-modal inputs including PDFs, audio, and video. Developers also have the option to completely disable network access to ensure the security of sensitive private data.
Google indicated that these two agents share the same underlying research framework as NotebookLM and Google Search, and will be further integrated into the enterprise market through Google Cloud. As self-research agents advance into the "long-term reasoning" era, their role in professional analysis is evolving from a mere information processor to an autonomous, planning-capable expert for deep analysis.
Related article
Anthropic Expands Claude AI Coding Tools to Japan in Push for Overseas Growth
Anthropic, a prominent U.S. artificial intelligence firm, is intensifying its global outreach. On Wednesday, the company hosted a major developer gathering, "Code with Claude," in Tokyo, drawing close to 500 software engineers. This initiative seeks
India’s Software Giant Cuts Hiring, Promises No Layoffs as AI Agents Scale Up
As artificial intelligence reshapes traditional labor-intensive business models, Tata Consultancy Services (TCS), a premier Indian software outsourcing firm, has unveiled its strategic response. During Tuesday’s annual general meeting, the chairman o
China Locks AI Models During Gaokam to Block Instant Homework Help
With the 2026 Gaokao fast approaching, rumors regarding the suspension of AI tools during the exam period have ignited intense online debate. In response to public concern, major AI platforms and educational apps have clarified their stance: rather t
Related Special Topic Recommendations
Comments (0)
0/500
Google recently unveiled two new self-research agents built on the Gemini 3.1 Pro architecture: Deep Research and Deep Research Max. These tools are now accessible for public preview via the paid tier of the Gemini API. They are designed to fully automate intricate research workflows, signaling a move for AI agents from basic web queries toward "long-term reasoning" models capable of deep analysis.

The standard Deep Research agent emphasizes efficiency and low latency, making it ideal for real-time conversational applications that need quick responses. In contrast, Deep Research Max focuses on depth of investigation, utilizing extended processing time for multi-step reasoning and iteration, primarily aimed at asynchronous backend tasks like due diligence reports. Technically, this new iteration introduces support for the Model Context Protocol (MCP), enabling agents to pull data from the open web as well as private databases containing financial or market information. Furthermore, the agents now feature native visualization tools, allowing them to directly generate charts and infographics in HTML format.

In benchmark performance tests, Google reports that Deep Research Max demonstrates substantial gains in retrieval and reasoning tasks compared to previous models. However, industry analysts note that comparisons with the OpenAI GPT-5.4 series and Anthropic's Opus4.6 can be influenced by testing methodologies and should be viewed with caution. A notable addition is a collaborative planning feature, supporting multi-modal inputs including PDFs, audio, and video. Developers also have the option to completely disable network access to ensure the security of sensitive private data.
Google indicated that these two agents share the same underlying research framework as NotebookLM and Google Search, and will be further integrated into the enterprise market through Google Cloud. As self-research agents advance into the "long-term reasoning" era, their role in professional analysis is evolving from a mere information processor to an autonomous, planning-capable expert for deep analysis.
Anthropic Expands Claude AI Coding Tools to Japan in Push for Overseas Growth
Anthropic, a prominent U.S. artificial intelligence firm, is intensifying its global outreach. On Wednesday, the company hosted a major developer gathering, "Code with Claude," in Tokyo, drawing close to 500 software engineers. This initiative seeks
India’s Software Giant Cuts Hiring, Promises No Layoffs as AI Agents Scale Up
As artificial intelligence reshapes traditional labor-intensive business models, Tata Consultancy Services (TCS), a premier Indian software outsourcing firm, has unveiled its strategic response. During Tuesday’s annual general meeting, the chairman o
China Locks AI Models During Gaokam to Block Instant Homework Help
With the 2026 Gaokao fast approaching, rumors regarding the suspension of AI tools during the exam period have ignited intense online debate. In response to public concern, major AI platforms and educational apps have clarified their stance: rather t











