Deep Research
Research API that allows you to build deep research into your own applications”
Introducing Deep Research (Alpha)
The /deep-research
endpoint enables AI-powered deep research and analysis on any topic. Simply provide a research query, and Firecrawl will autonomously explore the web, gather relevant information, and synthesize findings into comprehensive insights.
Building with Deep Research
Deep Research works by:
- Analyzing your query to identify key research areas
- Iteratively searching and exploring relevant web content
- Synthesizing information from multiple sources
- Providing structured findings with source attribution
Firecrawl provides structured results that enable you to build powerful applications:
- Activities: Detailed timeline of research steps and findings
- Sources: Curated list of relevant URLs with titles and descriptions
- Final Analysis: Comprehensive synthesis of key insights and conclusions
- Progress Tracking: Real-time status updates on research depth and completion
Example Usage
Key Parameters:
- query: The research topic or question you want to investigate
- maxDepth (Optional): Maximum number of research iterations (1-10, default: 7)
- timeLimit (Optional): Time limit in seconds (30-300, default: 270)
- maxUrls (Optional): Maximum number of URLs to analyze (1-1000, default: 20)
See API Reference for more details.
Response
Monitoring Research Progress
Deep Research jobs run asynchronously. You can monitor progress and receive real-time updates:
Research Activities
The data response includes:
- activities: List of research activities with the following properties:
type
: Type of activity (‘search’, ‘extract’, ‘analyze’, ‘reasoning’, ‘synthesis’, ‘thought’)status
: Activity status (‘processing’, ‘complete’, ‘error’)message
: Description of the activity or findingtimestamp
: ISO timestamp of when the activity occurreddepth
: Current research depth level- sources: Referenced URLs with titles and descriptions
title
: Title of the sourcedescription
: Description of the sourceurl
: URL of the sourceicon
: Icon of the source- finalAnalysis: Comprehensive analysis (when completed)
Status Examples
In Progress
Completed
JSON Output
You can now specify the JSON output format by setting the formats
parameter to json
. Set the jsonOptions
parameter to specify the schema for the JSON output.
Customize even further
You can also specify a systemPrompt
and an analysisPrompt
to customize the agentic process and the final analysis, respectively.
Models
While in Alpha we use a combination of small models to explore the web and synthesize information. That way we can keep the cost low and the research fast. But this can result in the synthesis not being very long and detailed. If you’d like us to offer more powerful models, let us know at help@firecrawl.dev.
Known Limitations (Alpha)
-
Research Scope
Best suited for topics with publicly available information. May not access paywalled or private content. -
Time Constraints
Research jobs are limited to 10 minutes maximum to ensure reasonable response times. -
Source Verification
While sources are provided, manual verification of critical information is recommended. -
Alpha State
As an Alpha feature, the research methodology and output format may evolve based on feedback.
Billing and Usage
Billing is done based on the number of urls
analyzed. Each url
is 1 credit. You can specify the max number of urls to analyze with the maxUrls
parameter.
Have feedback or need help? Email help@firecrawl.dev.