- Google just launched upgraded Deep Research and Deep Research Max agents inside the Gemini API, powered by Gemini 3.1 Pro.
- These agents handle complex multi-step tasks including file analysis, web searches, and creating fully cited reports with built-in charts and visualizations.
- Deep Research is built for speed and quick applications while Deep Research Max runs deeper overnight jobs like due diligence or market analysis.
- Developers now get multimodal inputs, collaborative planning, and native integrations with tools like Google Search, all available in paid tiers starting at roughly $1 to $3 per task.
- Early tests show strong benchmark performance, making these agents especially useful for finance, analytics, research, and business intelligence workflows.
Google Launches Upgraded Deep Research Agents in Gemini API
I came across this new release from Google and it feels like a real game changer for developers and researchers. The upgraded Deep Research and Deep Research Max agents are now live in the Gemini API. They bring serious multi-step reasoning capabilities that go way beyond simple chat responses.
These agents are powered by the latest Gemini 3.1 Pro model. They can analyze uploaded files, run web searches, and produce complete reports that include proper citations and native visualizations. This means you get polished outputs ready for professional use without extra manual work.
The two versions serve different needs. Deep Research focuses on speed so you can build quick applications that deliver fast results. Deep Research Max on the other hand dives much deeper and is designed for longer, more thorough jobs that can run overnight.
I see this as a smart way to give developers more control. The new agents support multimodal inputs so you can feed them text, images, PDFs, and spreadsheets all at once. They also include collaborative planning features that let multiple agents work together on a single project.
Here is a quick breakdown of the main capabilities:
- Automatic web searching with real-time results
- File analysis and data extraction
- Cited report generation with charts
- Multimodal input support
- Tool integrations including Google Search
To make the differences easier to understand, here is a simple comparison:
| Feature | Deep Research | Deep Research Max |
|---|---|---|
| Main focus | Speed and quick tasks | Depth and thorough analysis |
| Typical use case | Fast research apps | Overnight due diligence |
| Processing time | Minutes | Hours or overnight |
| Output detail level | Balanced | Maximum depth and citations |
| Best for | Daily workflows | Complex finance or analytics |
Pricing is straightforward for paid tiers. Tasks start around $1 to $3 depending on complexity and length. This makes the agents accessible for both small projects and enterprise level work.
Early testers already report impressive benchmark results. The agents perform especially well in areas like financial modeling, market research, and competitive intelligence. Many are now integrating them directly into existing business tools.
The launch also opens new possibilities for automation. Teams can now set up agents that handle entire research pipelines without constant supervision.
What Changes Next
This release will change how professionals approach research heavy work. Routine analysis tasks will move to the background while humans focus on higher level strategy and decision making. Small businesses and solo analysts will gain access to tools that were previously only available to large organizations with big research teams.
Finance professionals and data analysts should start experimenting with these agents right away. Test them on real datasets and workflows to see where they fit best. The earlier you build familiarity, the bigger the advantage you will have when usage scales across your industry.
Overall I expect this to accelerate AI adoption in knowledge work. The gap between simple chatbots and truly useful autonomous agents just got a lot smaller.
FAQs
What is the difference between Deep Research and Deep Research Max?
Deep Research prioritizes speed for quick tasks while Deep Research Max handles longer, more detailed analysis that can run overnight with deeper reasoning.
How much does it cost to use these agents?
Pricing starts in paid tiers at roughly $1 to $3 per task depending on complexity, length, and whether you choose the standard or Max version.
Can these agents work with my existing files and tools?
Yes. They support multimodal inputs, file uploads, and native integrations with services like Google Search for seamless workflow handling.
Who will benefit most from these new agents?
Finance teams, market researchers, analysts, and developers building research focused applications will see the biggest gains in productivity and output quality.
Are these agents available to everyone right now?
They are available now through the Gemini API for developers in paid tiers. Full rollout to consumer Gemini users may follow based on early feedback.