Trending Now - Google Launches Deep Research Agents for Enhanced Data Search

WHO:

Techmeme - Google has launched two new research agents: Deep Research and Deep Research Max.

VentureBeat - Google CEO Sundar Pichai announced the updates on X.

VentureBeat - Logan Kilpatrick leads developer relations for Google’s AI efforts.

VentureBeat - Google is collaborating with FactSet, S&P, and PitchBook on MCP server designs.

WHAT:

VentureBeat - The Deep Research Max agent achieved a score of 93.3% on DeepSearchQA, up from 66.1% in December 2025, and 54.6% on Humanity's Last Exam, up from 46.4%.

VentureBeat - The new agents can accept multimodal inputs including PDFs, CSVs, images, audio, and video as grounding context.

VentureBeat - The Deep Research agents are available in public preview via paid tiers of the Gemini API, which was first introduced in December 2025.

VentureBeat - The Deep Research Max agent is designed for asynchronous, background workflows, allowing for exhaustive, fully sourced analyses to be completed overnight.

VentureBeat - Google disclosed that it is actively collaborating with FactSet, S&P, and PitchBook on their MCP server designs.

VentureBeat - The new agents can connect to arbitrary third-party data sources through the Model Context Protocol (MCP).

VentureBeat - The pricing for the Deep Research agent was confirmed to be $2 per million input tokens and $2 per million output tokens with a 1 million token context window.

VentureBeat - The Deep Research agent is now optimized for low-latency, interactive use cases, delivering significantly reduced latency and cost at higher quality levels compared to its predecessor.

VentureBeat - Deep Research Max is designed for asynchronous, background workflows, allowing for exhaustive, fully sourced analyses to be completed overnight.

VentureBeat - The new agents generate high-quality charts and infographics inline within their reports, rendered in HTML or Google’s Nano Banana format, addressing previous limitations of text-only outputs.

VentureBeat - Google disclosed that it is actively collaborating with Axiom Bio to enhance research depth across biomedical literature using Deep Research.

Techmeme - The Deep Research agent replaces the preview agent released in December 2025.

VentureBeat - The new agents can connect to arbitrary third-party data sources through the Model Context Protocol (MCP).

VentureBeat - The new agents generate high-quality charts and infographics inline within their reports, rendered in HTML or Google’s Nano Banana format.

VentureBeat - Deep Research Max achieved a score of 93.3% on DeepSearchQA, up from 66.1% in December 2025, and 54.6% on Humanity's Last Exam, up from 46.4%.

Techmeme - The new agents are built with Gemini 3.1 Pro and bring MCP support, native visualizations, and improved analytical quality.

VentureBeat - Deep Research and Deep Research Max allow developers to combine open web data with proprietary enterprise information through a single API call.

VentureBeat - Deep Research Max achieved 93.3% on DeepSearchQA and 54.6% on Humanity's Last Exam.

VentureBeat - The agents generate high-quality charts and infographics inline within their reports.

VentureBeat - The pricing for the Deep Research agent was $2 per million input tokens and $2 per million output tokens with a 1 million token context window.

WHERE:

VentureBeat - The agents are available via the Gemini API, which was first introduced in December 2025.

WHEN:

VentureBeat - The Deep Research agents were launched on April 21st, 2026.

VentureBeat - The Gemini 3.1 Pro model was released on February 19th, 2026.

Techmeme - The Deep Research agents were launched on April 21st, 2026.

VentureBeat - The Gemini 3.1 Pro model was released on February 19th, 2026.

VentureBeat - Deep Research was first introduced as a consumer feature in December 2024.

WHY:

VentureBeat - Google aims to position its AI infrastructure as essential for enterprise research workflows in high-stakes industries like finance and life sciences, addressing the gap between open internet data and organizational needs.

VentureBeat - The launch represents Google’s effort to position its AI infrastructure as essential for enterprise research workflows in high-stakes industries like finance and life sciences.

VentureBeat - The addition of Model Context Protocol support allows Deep Research to securely query private databases and internal document repositories without requiring sensitive information to leave its source environment.

VentureBeat - The addition of MCP support allows Deep Research to securely query private databases and internal document repositories without requiring sensitive information to leave its source environment.

VentureBeat - The launch represents Google’s effort to position its AI infrastructure as essential for enterprise research workflows in high-stakes industries like finance and life sciences.

VentureBeat - The addition of MCP support allows Deep Research to securely query private databases and internal document repositories.

HOW:

VentureBeat - Developers can run Deep Research with Google Search, remote MCP servers, URL Context, Code Execution, and File Search simultaneously, or turn off web access entirely to search exclusively over custom data.

VentureBeat - Developers can now run Deep Research with Google Search, remote MCP servers, URL Context, Code Execution, and File Search simultaneously, or turn off web access entirely to search exclusively over custom data.

VentureBeat - Deep Research is optimized for low-latency, interactive use cases, while Deep Research Max is designed for asynchronous, background workflows.

VentureBeat - The agents can accept multimodal inputs including PDFs, CSVs, images, audio, and video as grounding context.

(UPDATED April 22, 2026 19:26 UTC)

Google's Deep Research Max agent has made significant strides, achieving a score of 93.3% on DeepSearchQA, a notable increase from 66.1% in December 20th25, and scoring 54.6% on Humanity's Last Exam, up from 46.4%. The new agents are designed to accept multimodal inputs, including PDFs, CSVs, images, audio, and video, and are now available in public preview through paid tiers of the Gemini API. This development aims to enhance AI infrastructure for enterprise research workflows in high-stakes industries like finance and life sciences, allowing developers to run Deep Research with various tools or limit searches to custom data.

(UPDATED April 22, 2026 15:22 UTC)

on Tuesday, Google launched its Deep Research Max agent, designed for asynchronous, background workflows that enable comprehensive analyses to be completed overnight. This new agent can connect to various third-party data sources through the Model Context Protocol, and it will cost $2 per million input and output tokens, with a context window of one million tokens. Google is also collaborating with FactSet, S&P, and PitchBook on their MCP server designs, aiming to solidify its AI infrastructure as crucial for enterprise research in high-stakes sectors like finance and life sciences.

(UPDATED April 22, 2026 13:20 UTC)

Google has announced significant upgrades to its Deep Research agent, which is now optimized for low-latency, interactive use cases, offering improved performance and lower costs compared to earlier versions. The new Deep Research Max is tailored for asynchronous workflows, enabling comprehensive analyses to be completed overnight, while also generating high-quality charts and infographics directly within reports. Additionally, Google is collaborating with Axiom Bio to deepen research capabilities in biomedical literature, and the latest updates allow developers to securely query private databases without exposing sensitive information.

(UPDATED April 22, 2026 01:25 UTC)

Techmeme reports that the Deep Research agent has officially replaced the preview agent released in December 20th25, introducing new capabilities that allow it to connect to various third-party data sources through the Model Context Protocol. This upgrade not only enhances its ability to generate high-quality charts and infographics inline within reports but also significantly improves its performance, with Deep Research Max achieving a score of 93.3% on DeepSearchQA, a notable increase from 66.1% just a few months ago. Additionally, developers can now run Deep Research in conjunction with Google Search and other remote servers or opt to disable web access to focus solely on custom data searches.

(UPDATED April 21, 2026 23:15 UTC)

Google has launched two new research agents, Deep Research and Deep Research Max, which were announced by CEO Sundar Pichai on X. Built with the Gemini 3.1 Pro model, these agents allow developers to combine open web data with proprietary enterprise information through a single API call, and they generate high-quality charts and infographics inline within their reports. The launch on Tuesday, marks a significant step in Google's efforts to enhance its AI infrastructure for enterprise research workflows, particularly in high-stakes industries like finance and life sciences.