Extract API

Updated Feb 2026

The Extract API (/v1/extract) gathers structured data from one or more URLs using LLM-powered extraction. It handles crawling, parsing, and collating datasets -- distinct from the JSON mode on scrape which operates on a single page.

Extract vs Scrape + JSON Mode

Scrape + JSON Mode (/v2/scrape): Extract structured data from one URL during a scrape. See LLM Extract.
Extract API (/v1/extract): Extract structured data across multiple URLs, supports async operations, wildcards, web search enrichment, and the FIRE-1 agent for complex interactions.

Overview

Rate Limit: 10 requests per minute

Extraction Modes

Mode	Description	When to Use
Schema-based	Provide a JSON Schema for rigid, typed output	You know the exact structure you need
Prompt-only	Let the LLM determine output structure	Exploratory extraction, flexible needs
Schema + Prompt	Schema for structure, prompt for guidance	Maximum control over extraction
No URLs (Alpha)	Extract using only a prompt, no URLs	Let the system find relevant data
FIRE-1 Agent	Browser automation for complex pages	Paginated data, dynamic elements, forms

Parameters

Parameter	Type	Required	Description
`urls`	string[]	No*	Target URLs; supports wildcards (e.g., `https://example.com/*`)
`prompt`	string	Yes	Natural language description of the data to extract
`schema`	object	No	JSON Schema defining the output structure
`enableWebSearch`	boolean	No	Follow external links for enriched results
`agent`	object	No	Agent configuration (e.g., `{"model": "FIRE-1"}`)

*URLs are optional in Alpha mode (prompt-only extraction).

URL Patterns

python

# Single URL
urls = ["https://example.com/pricing"]

# Multiple specific URLs
urls = [
    "https://example.com/pricing",
    "https://example.com/features",
    "https://example.com/about"
]

# Wildcard (experimental) — matches all pages on the domain
urls = ["https://example.com/*"]

Wildcard Support

Wildcard URL patterns (/*) are experimental. Coverage of large sites may be incomplete, and results can vary on dynamic sites.

Basic Usage

Schema-Based Extraction

Python

python

from firecrawl import FirecrawlApp

app = FirecrawlApp(api_key="fc-YOUR-API-KEY")

result = app.extract(
    urls=["https://firecrawl.dev/pricing"],
    prompt="Extract all pricing plans with their names, prices, and included features",
    schema={
        "type": "object",
        "properties": {
            "plans": {
                "type": "array",
                "items": {
                    "type": "object",
                    "properties": {
                        "name": {"type": "string"},
                        "monthly_price": {"type": "string"},
                        "annual_price": {"type": "string"},
                        "features": {
                            "type": "array",
                            "items": {"type": "string"}
                        }
                    },
                    "required": ["name", "monthly_price"]
                }
            }
        },
        "required": ["plans"]
    }
)

for plan in result['data']['plans']:
    print(f"{plan['name']}: {plan['monthly_price']}/mo")

Node.js

javascript

import FirecrawlApp from '@mendable/firecrawl-js';

const app = new FirecrawlApp({ apiKey: "fc-YOUR-API-KEY" });

const result = await app.extract({
  urls: ["https://firecrawl.dev/pricing"],
  prompt: "Extract all pricing plans with their names, prices, and included features",
  schema: {
    type: "object",
    properties: {
      plans: {
        type: "array",
        items: {
          type: "object",
          properties: {
            name: { type: "string" },
            monthly_price: { type: "string" },
            annual_price: { type: "string" },
            features: { type: "array", items: { type: "string" } }
          },
          required: ["name", "monthly_price"]
        }
      }
    },
    required: ["plans"]
  }
});

console.log(result.data);

cURL

bash

curl -X POST https://api.firecrawl.dev/v1/extract \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer fc-YOUR-API-KEY' \
  -d '{
    "urls": ["https://firecrawl.dev/pricing"],
    "prompt": "Extract all pricing plans with their names, prices, and included features",
    "schema": {
      "type": "object",
      "properties": {
        "plans": {
          "type": "array",
          "items": {
            "type": "object",
            "properties": {
              "name": {"type": "string"},
              "monthly_price": {"type": "string"},
              "features": {"type": "array", "items": {"type": "string"}}
            },
            "required": ["name", "monthly_price"]
          }
        }
      },
      "required": ["plans"]
    }
  }'

Prompt-Only Extraction

Let the LLM decide the output structure:

python

result = app.extract(
    urls=["https://firecrawl.dev"],
    prompt="Extract all information about the company's products, pricing, and team"
)

print(result['data'])
# Structure determined by the LLM based on page content

javascript

const result = await app.extract({
  urls: ["https://firecrawl.dev"],
  prompt: "Extract all information about the company's products, pricing, and team"
});

With Web Search Enrichment

Follow external links to gather additional context:

python

result = app.extract(
    urls=["https://example.com/company"],
    prompt="Extract company information and recent funding details",
    schema={
        "type": "object",
        "properties": {
            "name": {"type": "string"},
            "founded": {"type": "string"},
            "funding": {"type": "string"},
            "investors": {"type": "array", "items": {"type": "string"}}
        }
    },
    enableWebSearch=True  # Follow links for enrichment
)

With FIRE-1 Agent

For pages requiring browser interaction:

python

result = app.extract(
    urls=["https://example.com/products"],
    prompt="Navigate through all pagination pages and extract every product name, price, and rating",
    schema={
        "type": "object",
        "properties": {
            "products": {
                "type": "array",
                "items": {
                    "type": "object",
                    "properties": {
                        "name": {"type": "string"},
                        "price": {"type": "string"},
                        "rating": {"type": "number"}
                    }
                }
            }
        }
    },
    agent={"model": "FIRE-1"}
)

See FIRE-1 Agent Model for full documentation.

Async Operations

For long-running extractions, use the start/poll pattern:

Python

python

# Start the extraction job (returns immediately)
job = app.start_extract(
    urls=["https://example.com/*"],
    prompt="Extract all product information from the site",
    schema={...}
)

print(f"Job ID: {job['id']}")

# Poll for status
import time

while True:
    status = app.get_extract_status(job['id'])
    print(f"Status: {status['status']}")

    if status['status'] == 'completed':
        print(status['data'])
        break
    elif status['status'] == 'failed':
        print(f"Error: {status.get('error')}")
        break

    time.sleep(5)

Node.js

javascript

// Start the extraction job
const job = await app.startExtract({
  urls: ["https://example.com/*"],
  prompt: "Extract all product information from the site",
  schema: {...}
});

console.log(`Job ID: ${job.id}`);

// Poll for status
let status;
do {
  await new Promise(resolve => setTimeout(resolve, 5000));
  status = await app.getExtractStatus(job.id);
  console.log(`Status: ${status.status}`);
} while (status.status === 'processing');

if (status.status === 'completed') {
  console.log(status.data);
}

cURL

bash

# Start extraction
curl -X POST https://api.firecrawl.dev/v1/extract \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer fc-YOUR-API-KEY' \
  -d '{
    "urls": ["https://example.com/*"],
    "prompt": "Extract all product information",
    "schema": {...}
  }'

# Response: {"success": true, "id": "extract-abc123"}

# Check status
curl -X GET https://api.firecrawl.dev/v1/extract/extract-abc123 \
  -H 'Authorization: Bearer fc-YOUR-API-KEY'

Job Status Values

Status	Description
`processing`	Job is running
`completed`	Results are ready
`failed`	Job encountered an error
`cancelled`	Job was cancelled

Results are accessible for 24 hours after completion.

Response Format

json

{
  "success": true,
  "data": {
    "plans": [
      {
        "name": "Free",
        "monthly_price": "$0",
        "features": ["500 credits", "5 agent runs/day"]
      },
      {
        "name": "Starter",
        "monthly_price": "$19",
        "features": ["3,000 credits", "Batch scrape", "Priority support"]
      }
    ]
  }
}

Extract vs Agent API

Feature	Extract (`/v1/extract`)	Agent (`/v2/agent`)
Purpose	Structured extraction from known URLs	Autonomous research
URLs	Required (or wildcard)	Optional
Web Search	Optional (`enableWebSearch`)	Built-in
Browser Agent	FIRE-1 (optional)	Spark-1 Mini/Pro
Output	Schema-defined JSON	Schema or unstructured
Best For	Targeted data collection	Open-ended research

When to Use Which

Known URLs, need structured data --> Extract API
Don't know where the data is --> Agent API
Single page, simple extraction --> Scrape + JSON mode
Complex page interactions needed --> Extract API + FIRE-1

Pricing

Extract API uses a unified credit system where each credit equals approximately 15 tokens. Costs vary based on:

Number of URLs processed
Prompt complexity
Schema size
Data volume in responses
FIRE-1 agent usage (if enabled)

See Pricing for full details.

Limitations (Beta)

The Extract API is in beta. Current limitations:

Large sites: Full coverage of massive sites with wildcards may be incomplete
Complex queries: Highly complex logical queries may produce inconsistent results
Dynamic sites: Results may vary on frequently updated or JavaScript-heavy sites
Rate limit: 10 requests per minute

LLM Extract (JSON Mode) -- Single-page extraction on scrape
FIRE-1 Agent Model -- Browser automation agent
Agent API -- Autonomous research with Spark-1
AI Models -- Model comparison and selection
Batch Scrape -- Bulk URL scraping

Next: Features Overview -->

Extract API ​

Overview ​

Extraction Modes ​

Parameters ​

URL Patterns ​

Basic Usage ​

Schema-Based Extraction ​

Python ​

Node.js ​

cURL ​

Prompt-Only Extraction ​

With Web Search Enrichment ​

With FIRE-1 Agent ​

Async Operations ​

Python ​

Node.js ​

cURL ​

Job Status Values ​

Response Format ​

Extract vs Agent API ​

When to Use Which ​

Pricing ​

Limitations (Beta) ​

Related Pages ​

Extract API

Overview

Extraction Modes

Parameters

URL Patterns

Basic Usage

Schema-Based Extraction

Python

Node.js

cURL

Prompt-Only Extraction

With Web Search Enrichment

With FIRE-1 Agent

Async Operations

Python

Node.js

cURL

Job Status Values

Response Format

Extract vs Agent API

When to Use Which

Pricing

Limitations (Beta)

Related Pages