curl -X POST 'https://api.firecrawl.dev/v1/deep-research' \
  -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "query": "string",
  "maxDepth": 7,
  "timeLimit": 300,
  "maxUrls": 20,
  "analysisPrompt": "string",
  "systemPrompt": "string",
  "formats": [
    [
      "markdown"
    ]
  ],
  "jsonOptions": {
    "schema": {},
    "systemPrompt": "string",
    "prompt": "string"
  }
}'

# POST /deep-research (Firecrawl API)
Base URL: https://api.firecrawl.dev/v1
Auth: Authorization: Bearer <FIRECRAWL_API_KEY>
Last verified: 2026-09-03

> Start a deep research operation on a query

## Critical Gotchas
* The `formats` field controls the output shape and defaults to `["markdown"]`. Omitting it does not return HTML — request `["markdown","html"]` explicitly.
* `/crawl` is asynchronous: it returns a job `id`, and results are collected by polling `/crawl/{id}` until `status` is `completed`.

## cURL
curl -X POST 'https://api.firecrawl.dev/v1/deep-research' \
  -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"query":"string","maxDepth":7}'

## Parameters
| Name | In | Type | Required | Description |
| :--- | :--- | :--- | :--- | :--- |
| query | body | string | Yes | The query to research |
| maxDepth | body | integer | No | Maximum depth of research iterations |
| timeLimit | body | integer | No | Time limit in seconds |

## Response (200 OK)
{"success":true,"id":"00000000-0000-4000-8000-000000000000"}

---
Need an API key? Get one at https://www.firecrawl.dev/signin/signup

POST

/deep-research

Firecrawl API

Start a deep research operation on a query

Base URL: https://api.firecrawl.dev/v1
Auth: Authorization: Bearer <FIRECRAWL_API_KEY>
Last verified: 2026-09-03 · upstream hash matched

Actions

Agents: curl -H "Accept: text/markdown" this URL
→ 219 tokens · Vary: Accept

Critical gotchas

The formats field controls the output shape and defaults to ["markdown"]. Omitting it does not return HTML — request ["markdown","html"] explicitly.

/crawl is asynchronous: it returns a job id, and results are collected by polling /crawl/{id} until status is completed.

cURL

curl -X POST 'https://api.firecrawl.dev/v1/deep-research' \  -H "Authorization: Bearer $FIRECRAWL_API_KEY" \  -H 'Content-Type: application/json' \  -d '{  "query": "string",  "maxDepth": 7,  "timeLimit": 300,  "maxUrls": 20,  "analysisPrompt": "string",  "systemPrompt": "string",  "formats": [    [      "markdown"    ]  ],  "jsonOptions": {    "schema": {},    "systemPrompt": "string",    "prompt": "string"  }}'

Get a free Firecrawl API key → sponsored

Parameters

Name	In	Type	Required	Description
query	body	string	Yes	The query to research
maxDepth	body	integer	No	Maximum depth of research iterations
timeLimit	body	integer	No	Time limit in seconds
maxUrls	body	integer	No	Maximum number of URLs to analyze
analysisPrompt	body	string	No	The prompt to use for the final analysis. Useful to format the final analysis markdown in a specific way.
systemPrompt	body	string	No	The system prompt to use for the research agent. Useful to steer the research agent to a specific direction.
formats	body	string[]	No	Request body field formats.
jsonOptions	body	object	No	Options for JSON output

Response 200 OK

{
  "success": true,
  "id": "00000000-0000-4000-8000-000000000000"
}