Firecrawl logo
Web Scraping · Web Research Browsing

Firecrawl

Firecrawl is an open-source context API that turns any web page into clean Markdown or structured JSON, and can search the live web or click through it. Built for AI agents, RAG pipelines and research tooling.

Active Free plan Freemium API available 13+ Verified by Guidaio
Overview

What is Firecrawl?

Firecrawl describes itself as the context API to search, scrape, and interact with the web at scale: one interface that turns live websites into clean, LLM-ready data. Where a classic scraper hands back raw HTML, Firecrawl returns Markdown, structured JSON, screenshots or semantic text, with navigation, footers and advertising stripped out before the content ever reaches a model.

Three capabilities sit at the core. Search queries the web and returns the full content of each result rather than a list of links. Scrape converts a single URL into whichever output format you ask for. Interact opens a cloud browser sandbox where an agent can click, scroll, type, wait, press keys and capture the screen, driven either from code or in plain language. Around them sit Crawl for whole sites, Map for URL discovery, Monitor for change detection, Parse for PDFs and DOCX files, an Agent mode still in preview, and a pre-built Developer Index for cases where latency matters more than freshness.

The publisher is SideGuide Technologies, Inc., trading as Firecrawl, a Delaware corporation run from San Francisco and part of Y Combinator's S22 batch. The same team previously built Mendable, an AI chat layer over documentation used by Snapchat, MongoDB and DoorDash, and concluded that getting clean web data into AI systems was the real bottleneck. The core is open source on GitHub, where the repository shows 176,900 stars; the hosted version adds Fire-Engine, the proprietary infrastructure handling proxies and rendering, plus the Interact capability, an analytics dashboard and the unified API.

The scale figures come from the publisher and should be read as claims rather than audited facts: more than 5 billion requests served, over 1.25 million registered developers, 150,000+ companies, coverage of 96% of the web including JavaScript-heavy pages, a P95 latency of 3.4 seconds and 93% fewer input tokens thanks to content cleaning. Named customers include Zapier, Replit, Lovable, Gamma, Cognism and Sierra.

Access is deliberately agent-friendly: Python and Node.js SDKs, cURL, a CLI that wires an editor up in one command, an official MCP server, skills for Claude Code and Codex, and no-code connectors through n8n, Zapier, Make and Dify. Billing runs on monthly credits per plan, topped up pay-as-you-go in $5 increments.

What it does

  • Scrape any URL into clean Markdown, structured JSON, a screenshot or semantic text
  • Search the web and get back the full content of each result, not just a list of links
  • Interact with live pages: click, scroll, type, wait and navigate dynamic content
  • Crawl an entire site page by page, or map its URLs in a single call
  • Parse PDFs, DOCX files and other documents into usable text
  • Monitor chosen pages and detect what changed since the last check
  • Query a pre-built Developer Index when speed matters more than freshness
Audience

When to use Firecrawl / When not to

A quick filter to help you decide if Firecrawl is the right fit.

When to use Firecrawl

  • Engineering teams building AI agents that need fresh, LLM-ready web context
  • Developers assembling RAG pipelines over documentation, product sites and public sources
  • Sales and growth teams running lead enrichment, competitive monitoring or price tracking
  • Market analysts and academic researchers who need deep research across many sources at once
  • Solo builders and side projects, covered by the permanent free tier and the entry-level Hobby plan

When not to use Firecrawl

  • Non-technical users looking for a point-and-click scraper: everything runs through code, a CLI or a third-party connector
  • Organizations that must keep data inside the EU, since hosting is United States only with no residency option
  • Anyone expecting a mobile app, as the product is API and web only
  • Buyers on a small budget who need SSO/SAML, SCIM, IP allowlisting or Zero Data Retention, all of which sit behind Enterprise
  • Teams that require a contractual support SLA, which is listed as unavailable from Free through Scale
Get started

How to use Firecrawl

A typical end-to-end flow, from setup to results.

  1. Sign up with no payment card required and collect the 1,000 free credits granted every month
  2. Copy the API key issued in the dashboard, recognizable by its fc- prefix
  3. Try a first request without writing any code in the public playground
  4. Or install an SDK, Python or Node.js, and call scrape on a single URL
  5. Read the result back as Markdown, JSON, a screenshot, semantic text or a summary
  6. Move on to Search, Crawl, Map, Monitor or Parse depending on the use case
  7. Use Interact to drive a page, for example scraping a shop and then sending a natural-language instruction to run a search on it
  8. Wire an editor or coding agent up in one command with npx -y firecrawl-cli@latest init --all --browser
  9. Or connect through the official MCP server, or through n8n, Zapier, Make and Dify for no-code workflows
  10. Enable pay-as-you-go in the billing settings and set a monthly ceiling in multiples of $5, or 0 to switch it off
Quick read

Pros & Cons

Pros

  • Open-source core that can be audited on GitHub or self-hosted, with 176,900 stars displayed
  • Permanent free plan of 1,000 credits a month, with no payment card requested
  • Credit pricing identical across plans, published in plain text as well as HTML, and failed requests are not billed
  • SOC 2 Type II, penetration testing, MFA, signed webhooks and PII redaction included from the free plan upward
  • Unusually wide and explicitly named integration ecosystem, from LangChain and LlamaIndex to n8n, Zapier, Make and around forty MCP clients
  • First-class agent access through an official MCP server, a one-command CLI, Claude Code and Codex skills and a machine-readable onboarding path
  • Self-service all the way to the Scale plan without a sales call, plus a standing annual discount of 16.7% (20% on Scale)

Cons

  • The legal paperwork lags the product: the terms of service were last revised on 5 November 2024 and the privacy policy on 26 December 2024, and both still describe a small business while the site now advertises SOC 2 Type II, Enterprise plans and 1.25 million developers
  • The license granted over user content is very broad: worldwide, irrevocable, sub-licensable and surviving termination
  • Hosting is United States only, with no European data-residency option and an explicit warning about the level of protection after transfer
  • No explicit GDPR compliance claim, no Article 27 EU representative, no standard contractual clauses, and no postal address published anywhere on the site
  • Zero Data Retention, SSO, SCIM and IP restrictions are Enterprise-only, and no DPA is available on the Free or Hobby plans
  • The HTML pricing page shows the annual-equivalent figures ($16, $83, $333, $599) by default, while the real monthly prices ($19, $99, $399, $749) appear only in the Markdown version of the page
  • Credits do not roll over on Hobby, Standard or Growth, and purchased credits are lost on cancellation
Pricing

Pricing & Plans

A permanent free plan is available, providing 1,000 credits per month without a payment card. The lowest paid entry point is the Hobby plan at USD 19 per month on monthly billing, or the equivalent of USD 16 per month on an annual commitment (USD 190 per year). The higher plans are Standard at USD 99 per month, Growth at USD 399 per month and Scale at USD 749 per month, with Enterprise priced on request. Additional usage may be purchased pay-as-you-go in USD 5 increments on paid plans. The annual discount of 16.7% on Hobby, Standard and Growth, and 20% on Scale, is a standing billing condition rather than a dated promotion.

Free
  • 1
  • 000 credits per month at USD 0
  • 2 concurrent requests
  • low rate limits and community support
  • enough for roughly 500 searches or 1
  • 000 pages
Standard
  • 100
  • 000 credits per month at USD 99 monthly (USD 83 per month annually)
  • 25 concurrent requests
  • standard support
  • a countersigned standard DPA and 2
  • 000 extra credits per USD 5 top-up
Growth
  • 500
  • 000 credits per month at USD 399 monthly (USD 333 per month annually)
  • 50 concurrent requests
  • priority support
  • a shared Slack channel and 2
  • 500 extra credits per USD 5 top-up
Scale
  • 1
  • 000
  • 000 credits per month at USD 749 monthly (USD 599 per month annually)
  • 100 concurrent requests
  • priority support
  • one month of credit rollover
  • invoice billing and 5
  • 000 extra credits per USD 5 top-up
Enterprise
  • custom credit volume and custom pricing
  • dedicated support with an SLA
  • volume discounts
  • Zero Data Retention
  • SSO and advanced security
  • SCIM
  • static egress IPs and custom analytics
Special offers — Annual commitment: 16.7% off Hobby, Standard and Growth, and 20% off Scale, presented on site as a 2 Months Free banner. This is a standing billing condition, not a dated promotion · Student program announced on a dedicated page, with conditions not published in the material collected · Agent mode in preview: 5 free runs per day before dynamic pricing applies · Research Index paper endpoints offered free across all categories · Ambassador program and Firestarters program mentioned in the site footer · No dated promotion and no discount code are advertised anywhere on the site
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how Firecrawl handles your data.

GDPR overview

Firecrawl never claims GDPR compliance in so many words: the acronym appears once in the terms of service, and only as an obligation placed on the user. The privacy policy nonetheless carries an EU Rights to Information section granting eight rights, namely information, access, rectification, erasure, restriction, portability, objection including profiling and direct marketing, and automated decision-making, and states that they are respected for all users regardless of citizenship. Requests go to help@firecrawl.com, and the right to complain to a national supervisory authority is acknowledged. Transfers are explicit: data leaves the home country for the United States, with a warning that protection there may be weaker. What is absent is just as clear, with no Article 27 representative, no named DPO and no standard contractual clauses. A countersigned standard DPA is offered from the Standard plan upward.

Who owns the data?

The terms of service state that content and intellectual property posted by users belongs to the user who posted it. That ownership comes with a wide grant, however: by submitting content or feedback you give the publisher a worldwide, irrevocable, non-exclusive, royalty-free license to use, reproduce, modify, publish, translate and distribute it in any existing or future medium, together with the right to sub-license those rights, and the grant survives termination. Ownership of the Services themselves stays with SideGuide Technologies; you hold rights only over what you generate. Payment cards are not kept in-house but held by Stripe. The terms were last revised on 5 November 2024 and the privacy policy on 26 December 2024.

Reuse rights

Nothing in the terms limits what you do with the content you submit or the data you pull back: you keep ownership and need no permission to reuse it, while the contract pushes responsibility for lawful use, including GDPR and CCPA compliance, onto you. On the publisher's side, the privacy policy lists service delivery, caching and indexing, service, product and billing emails, site performance, personalization, payment processing, product improvement, marketing and advertising, with cookies tracking unauthenticated visitors. The named third-party recipients are Stripe for payments, PostHog for product analytics, Crisp for support chat and Vercel Analytics for marketing. Neither document says anything about training AI models on customer data, which is a silence rather than a commitment. Do Not Track signals are honored, and data is encrypted in transit over TLS, at rest only on request.

Data retention & training

Retention summary
Personal data is kept until the user asks in writing for it to be deleted. The privacy policy states plainly that no scheduled deletion policy currently exists, and that one may be introduced later. Erasure is granted to all users under the EU rights section, with requests sent to help@firecrawl.com; identity may be verified before a request is processed. Under the CCPA, a written request, up to twice per twelve-month period, obtains a summary of the data held, and deletion follows unless a legal retention obligation applies. Zero Data Retention for scraped content, meaning nothing is kept after the request, is restricted to Enterprise and to approved use cases. Payment cards are not stored in-house but held by Stripe. Several sections of the terms of service survive termination.
Subprocessors disclosed
Yes
DPA available
Yes
GDPR contact

Hosting summary

The privacy policy is unambiguous on jurisdiction: the business is operated from the United States, the servers are located in the United States, and that is where user data and information are stored. As far as the publisher is aware, the third parties it shares information with are American as well. The document adds the usual caveat that, given the nature of internet communications, data may transit through other countries along the way, and it explicitly warns European users that their information leaves their home country for a jurisdiction that may not offer the same level of protection. No European data-residency option is offered at any plan level. Zero Data Retention, meaning no storage of scraped content after the request, is available only on Enterprise and only for approved use cases. Domain resolution points to 76.76.21.21, an anycast node operated within Amazon's AS16509 and located in the United States, which is consistent with the declared hosting.

Hosting countries
🇺🇸 United States
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting Firecrawl.

  • Pricing display: the HTML pricing page shows the annual-equivalent amounts by default, so a monthly subscription actually costs USD 19, 99, 399 or 749 rather than the 16, 83, 333 or 599 shown. The 2 Months Free banner describes a standing annual-commitment discount, not a dated promotion
  • Content rights: the license granted to the publisher over submitted content is irrevocable, sub-licensable and survives the end of the contract, so anything sent through the service should be reviewed with that in mind
  • Retention: the privacy policy, dated 26 December 2024, states that no scheduled deletion policy exists, meaning personal data is kept until you ask in writing for it to be removed
  • Jurisdiction: data is stored in the United States and the transfer out of your home country is explicitly acknowledged, with a warning that the level of protection may be lower than at home
  • Credits: purchased pay-as-you-go credits are lost on cancellation, and unused monthly credits do not roll over except on Scale and Enterprise
  • Compliance sits with you: the terms make the user responsible for respecting GDPR and CCPA when scraping, and for the lawfulness of what is collected, which is easy to overlook when extraction is this frictionless
  • Over-reliance: machine-cleaned content strips navigation, footers and ads, but it can also drop context and nuance, so feeding it straight into a model without spot-checking the source encourages confident, unverified conclusions
Setup

Setup & Integrations

Technical difficulty

Low to moderate, and clearly aimed at developers. Signing up takes seconds, needs no payment card and issues an fc- prefixed API key immediately, after which a first call fits into three lines of Python or a single cURL command. The public playground allows a test with no code at all, a single CLI command connects an editor or coding agent, and n8n, Zapier, Make and Dify cover no-code workflows. Beyond that first call, the product assumes familiarity with SDKs, API keys, concurrency limits and rate limits. Self-hosting the open-source core is possible but ships without Fire-Engine or Interact.

Deployment

Web appAPIPlugin

Integrations

LangChain LlamaIndex OpenAI Anthropic Claude Google Gemini LangGraph Vercel AI SDK Mastra N8n Zapier Make Dify Cursor Claude Code Windsurf Gemini CLI Cline Smithery Claude Desktop VS Code Qwen Coder OpenAI Codex Zed Amp Augment Code Roo Code Opencode JetBrains AI Assistant Kiro Trae Docker Amazon Q Developer CLI Warp Copilot Coding Agent LM Studio Visual Studio Crush BoltAI Rovo Dev CLI Zencoder Qodo Gen Perplexity Desktop Lovable Stripe
Company

Behind Firecrawl

Company name
SideGuide Technologies, Inc.
Founded
13/04/2024
Country of origin
🇺🇸 United States
UBO
INFORMATION_NOT_FOUND
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇺🇸 United States
Legal contact
Support contact

Fundraising

USD 14.5M Series A announced on 19 August 2025, led by Nexus Venture Partners
Investors named in the announcement: Y Combinator, extending its initial investment, Zapier, a customer that became a shareholder, Tobias Lutke (CEO of Shopify), Abhinav Asthana (CEO of Postman) and Matt McClure (founder of Mux), alongside unnamed angels and funds
Y Combinator S22 batch, the earliest backing disclosed by the company
Figures quoted at the time of the Series A: 350,000+ registered developers, 48,000 GitHub stars and 15x growth over the year
No other round is announced on the site, and cumulative totals circulating in the press are not published by the publisher

Social

Official links

Resources

All the official URLs gathered for verification and reference.

Compare

Alternatives

Tools that compete with or complement Firecrawl.

T TavilyE ExaA ApifyB Bright DataS ScrapingBeeS SteelB Browser UseB Brave Search APIH HyperbrowserP ParallelS ScrapflyJ Jina AIZ ZyteB BrowserbaseC Crawl4AIS Scrape.doS ScrapeGraphAIS ScraperAPIC Context.dev
FAQ

Frequently asked questions

What exactly is Firecrawl?
Firecrawl is a web context API for AI. A single interface lets you search the web, scrape a URL and interact with a page, and returns clean, LLM-ready data as Markdown, structured JSON, a screenshot or semantic text rather than raw HTML.
Is there a free plan?
Yes. The free plan grants 1,000 credits per month with no payment card required, which covers roughly 500 searches or 1,000 pages. It is a permanent tier rather than a time-limited trial, so it does not expire after a set number of days.
How much does the cheapest paid plan cost?
The Hobby plan costs USD 19 per month on monthly billing. Committing annually brings that down to the equivalent of USD 16 per month, or USD 190 per year. Note that the HTML pricing page displays the annual-equivalent figure by default.
How are credits counted?
Scrape, Crawl, Map and Monitor cost 1 credit per page, Search costs 2 credits per 10 results, and Interact costs 2 credits per browser minute. Structured JSON output adds 4 credits per page. Failed requests are not billed, since only successful requests are charged.
Is Firecrawl really open source?
The core is open source in the firecrawl/firecrawl repository on GitHub, which the publisher presents as one of the top 100 repositories on the platform. The hosted version adds Fire-Engine, the proprietary proxy and rendering infrastructure, the Interact capability and an analytics dashboard.
Does it work with AI agents and MCP?
Yes. Firecrawl ships an official MCP server, a CLI that connects editors and agents in a single command, and skills for Claude Code and Codex. Named integrations include LangChain, LlamaIndex, OpenAI, Anthropic Claude, Google Gemini, Cursor, Windsurf, n8n, Zapier, Make and Dify.
Where is my data hosted, and can I get a DPA?
The privacy policy states that the servers are located in the United States and that this is where data is stored, with no European residency option. A countersigned standard DPA is available from the Standard plan upward, and a custom one at Enterprise level.
Is there a mobile app, and what is the minimum age?
There is no iOS or Android app: Firecrawl is an API and web product only. The terms of service set the minimum age at 18, or 13 with the express consent of a parent or guardian.
Conclusion

Should you pick Firecrawl?

Firecrawl is best read as data infrastructure for AI rather than as one more scraping tool. The proposition holds together: a single API covering search, extraction and browser interaction, an open-source core that can be audited or self-hosted, a permanent free tier of 1,000 credits a month with no card required, credit pricing published transparently, and no charge for failed requests. Security features that many competitors gate behind enterprise contracts, including SOC 2 Type II, MFA, signed webhooks and PII redaction, are present from the free plan upward. The integration surface is unusually wide, and the self-service path runs all the way to the Scale plan without a sales call.

The reservations are legal and geographic more than technical. The terms of service date from November 2024 and the privacy policy from December 2024; both still portray a small business and have visibly fallen behind a product that now advertises SOC 2 Type II, Enterprise tiers and a countersigned DPA. The license users grant over submitted content is broad, being worldwide, irrevocable, sub-licensable and surviving termination. Data is hosted exclusively in the United States, with no European residency option, no explicit GDPR compliance claim, no Article 27 representative and no standard contractual clauses named. No postal address appears anywhere on the site, and the single human contact channel sits on a different domain from the product itself. The adoption figures put forward, 1.25 million developers, 150,000 companies and 5 billion requests, remain the publisher's own claims.

In short: for a developer or an engineering team building agents and RAG pipelines, Firecrawl is a strong, well-documented and generously free option. For an organization with strict European data-residency needs or a demanding legal review, the paperwork should be updated, and questioned, before anything is signed.