Google shipped two autonomous research agents on Gemini 3.1 Pro — Deep Research and Deep Research Max — and the headline is MCP. Anthropic’s protocol is now the connective tissue inside Google’s flagship agent surface.
What it does
Point it at a question. It runs long-horizon research across Google Search, your URLs, files, and any MCP server you connect — financial feeds, market data providers, internal knowledge bases. Output isn’t a wall of text. Charts and infographics render inline (HTML or Nano Banana). A research report actually looks like one.
93.3% on DeepSearchQA, up from 66.1% in December. 54.6% on Humanity’s Last Exam.
API and pricing
Public preview via Gemini API, paid tiers only. Two SKUs: deep-research-preview-04-2026 at ~$2/task, deep-research-max-preview-04-2026 at ~$5/task. Search grounding adds $1-$2 on top. Built on the Interactions API — wire it into your own product instead of forcing users into the Gemini app. Multimodal grounding accepts PDFs, CSVs, images, audio, video.
Why it matters
Google adopting MCP at the flagship level is a watershed for Anthropic’s protocol. It’s now the de-facto standard across frontier labs. Deep Research moves from chat answer to real multi-tool research workflow — and the Max tier is the first agent priced like a small consulting task instead of a chat reply.
You Might Also Like
- Google A2ui Agent to User Interface Finally a Standard way for ai Agents to Show you Things
- Agent Action Protocol aap the Missing Layer Above mcp That Actually Makes Agents Production Ready
- Notion mcp Scores Product Hunt 1 With 408 Votes the Moment Mainstream Saas Joined the ai Agent Race
- Addy Osmani Open Sources Agent Skills 19 Workflows That Make ai Agents Code Like Google Engineers
- Cloudflare Stripe Agent Provisioning Protocol ai Agents can now buy Domains and Ship Apps Without you
Top AI Product 
Leave a comment