Google released Deep Research Max, a substantially upgraded autonomous research agent built on Gemini 3.1 Pro, on April 21, 2026. It adds Model Context Protocol (MCP) integration so agents can tap proprietary data alongside the open web, natively generates charts and infographics inline with reports, and accepts PDFs, CSVs, images, audio, and video as grounding inputs. Google positions it for "asynchronous, background workflows" like overnight due-diligence runs, and says it consults significantly more sources than the December 2025 version with measurably higher expert evaluation win-rates.