For developers who want cited research reports inside their own apps; it's preview-only and token costs depend on how long it digs.
Kenning's take
What it does
The Deep Research agent in the Gemini API plans a research task, searches and reads sources over several minutes, and writes a detailed report with citations. Developers can plan with it, connect MCP tools, pass in documents and get charts. It runs in the background through the Interactions API and is in preview.
- Multi-step autonomous research
- Cited reports
- Collaborative planning
- MCP server connections
- Charts and visualisations
- Background execution with polling or streaming
Who it suits. Developers adding long-form research to their own apps.
Key facts
- Category
- Developer tools
- Part of
- Google Gemini
- Made by
- Based in
- United States
- How to get it
- Gemini API via the Interactions API only, in preview
- Works on
- API
- In the UK
- Yes
- API price
- Standard Gemini list rates for input, output and reasoning tokens, plus search and tool fees
FAQ
How do I use the Gemini Deep Research agent?
Through the Interactions API with background execution turned on; it is not available through generate_content.
How is it priced?
At standard Gemini token rates for all model use, plus the usual tool fees.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
Gemini Deep Research Max
Google's slower, more thorough research agent for developers, built on Gemini 3.1 Pro
Gemini CLI
Google's open-source AI agent that runs Gemini in the terminal for coding and other tasks
Gemini Computer Use
Gemini API tool that lets AI agents operate browsers, phones and desktops through screenshots
Gemini Enterprise Agent Platform
Google Cloud's platform for building and running AI agents, which replaced Vertex AI