Executive Summary
Google has launched a significantly more powerful version of its Gemini Deep Research agent, making its most advanced autonomous research capabilities available to developers for the first time through the new Interactions API. The agent, powered by a Gemini 3 Pro reasoning core, is optimized for complex, long-running synthesis tasks, combining web search with analysis of user-provided documents to generate detailed, citable reports. Alongside the agent, Google has open-sourced DeepSearchQA, a new benchmark designed to evaluate agents on intricate, multi-step information-seeking tasks.
Key Takeaways
* New Agent Release: A vastly improved Gemini Deep Research agent is now available, designed for in-depth research and report generation.
* Developer Access via API: Developers can now embed the agent's capabilities into their own applications using the new Interactions API, accessible with a Gemini API key from Google AI Studio.
* Core Technology: The agent utilizes a Gemini 3 Pro reasoning core and is trained with multi-step reinforcement learning to reduce hallucinations and improve accuracy in complex search tasks.
* Advanced Capabilities:
* Synthesizes information from both public web data and user-uploaded files (PDFs, CSVs, docs).
* Provides granular citations for all generated claims.
* Supports structured JSON schema outputs for easy integration into other applications.
* Allows users to control the final report's structure and formatting via prompting.
* New Open-Source Benchmark: Google has released DeepSearchQA, a new benchmark with 900 tasks to evaluate an agent's comprehensiveness and precision on complex research queries.
* Availability: The agent is available now via the Interactions API and is slated for future integration into Google Search, NotebookLM, Google Finance, the Gemini App, and Vertex AI for enterprise customers.
Strategic Importance
This release positions Google's agent technology as a foundational platform for developers, aiming to drive adoption in high-value enterprise verticals like finance and biotech. By also open-sourcing a new, complex benchmark, Google seeks to steer the industry's research and establish its architecture as the standard for capable AI research agents.