Skip to main content

Overview

The Deep Research API enables comprehensive, multi-layered research by automatically conducting iterative searches and analysis to generate detailed, well-sourced reports. This powerful tool goes beyond simple search results to provide thorough investigations of complex topics with AI-powered synthesis and analysis.
  • Conduct deep, recursive research with configurable depth and breadth
  • Automatic follow-up queries based on initial findings
  • Customizable research parameters for different use cases
  • High-quality output with proper citations and source attribution

API Endpoint

Quick Start

Javascript

Response

Examples

Simple Research

Advanced Research

Best Practices

Deep research is an agentic process that conducts multi-step research, which can take several minutes to complete comprehensive investigations. To optimize your experience and results:

Performance & Reliability

  1. Set appropriate timeouts: Deep research requires time for thorough analysis. Configure higher timeout values in your HTTP client to accommodate longer processing times.
  2. Control research scope: Use max_depth and max_breadth parameters to balance thoroughness with performance. Higher values provide more comprehensive results but take longer to complete.
  3. Manage token limits: Set max_output_tokens to control response size and target_output_tokens to guide the desired length of your research report.

Query Optimization

  1. Be specific and detailed: Well-formed, detailed queries produce more focused and relevant research. Instead of “climate change,” try “impact of climate change on Arctic ice melting rates 2020-2024.”
  2. Provide context: Include relevant background information, constraints, or specific aspects you want the research to focus on.
  3. Consider research depth: For complex topics requiring extensive investigation, allow for higher max_depth values to enable recursive research iterations.

Cost Management

  1. Monitor usage: Deep research can generate substantial token usage due to its comprehensive nature. Track your API consumption and adjust parameters accordingly.
  2. Balance breadth vs depth: Use max_breadth to control how many parallel research paths are explored at each level, helping manage both cost and processing time.
Find more information on Deep Research API here
Check out the Open Deep Research Framework here