On April 19, 2026, Google DeepMind announced a major step change for its autonomous research agents, introducing two new configurations — Deep Research and Deep Research Max — both built on Gemini 3.3 Pro and delivered through the Gemini API. The release builds on the original Gemini Deep Research agent that shipped to developers in December 2025 via the Interactions API, and it reframes deep research from a consumer feature into a foundation for enterprise agentic pipelines.
With Gemini 3.3 Pro integration, Deep Research is positioned for enterprise workflows across finance, life sciences and market research. A single API call now triggers an exhaustive research workflow that blends the open web with proprietary data streams to produce professional-grade, fully cited analyses. Google frames these reports as valuable standalone deliverables and as the context-gathering first step of larger multi-agent pipelines.
The release splits into two distinct configurations. Deep Research prioritizes speed and efficiency, replacing the December preview with significantly reduced latency and cost while raising quality — making it suitable for research experiences embedded directly in interactive user surfaces. Deep Research Max instead maximizes comprehensiveness and synthesis quality, leveraging extended test-time compute to iteratively reason, search and refine. Max is designed for asynchronous background workflows, such as a nightly cron job that produces exhaustive due-diligence reports for analyst teams to read by morning, and it demonstrates substantial improvements across industry-standard retrieval and reasoning benchmarks.
A central new capability is Model Context Protocol (MCP) support. Deep Research can now securely connect to arbitrary remote MCP servers, custom data streams and specialized professional repositories, transforming it from a web searcher into an autonomous agent that navigates gated data universes. Google announced collaborations with FactSet, S&P Global and PitchBook on MCP server designs so that shared customers can pipe premium financial data directly into Deep Research workflows.
The agents also gain native charts and infographics — a first for Deep Research in the Gemini API. The system generates high-quality visualizations inline using HTML or Nano Banana image generation, dynamically illustrating complex datasets within its analytical reports. Additional capabilities include collaborative planning, where users review, guide and refine the research plan before execution; extended tooling that combines Google Search, remote MCP servers, URL Context, Code Execution and File Search simultaneously (or disables web access entirely); multimodal research grounding across PDFs, CSVs, images, audio and video; and real-time streaming of intermediate reasoning steps with live thought summaries plus incremental text and image outputs.
On real-world performance, Google states that compared to the December 2025 release, Deep Research Max consults significantly more sources and surfaces critical nuances that were previously overlooked, drawing on authoritative sources such as SEC filings and open-access peer-reviewed journals to deliver rigorous factuality and expert-grade analysis. Both Deep Research and Deep Research Max became available immediately in public preview on paid tiers of the Gemini API, with availability for startups and enterprises on Google Cloud following shortly after.