New — Agent runs, webhooks and semantic search are live
v2 is live — faster crawls, semantic search

Power AI agents with
clean web data

One API to scrape, crawl and search any site. Clean markdown, structured JSON, and screenshots — no proxies, no headless browser babysitting.

500 free credits · No credit card required

from firecrawl import Firecrawl

app = Firecrawl(api_key="fc-YOUR_KEY")

doc = app.scrape(
    "https://firecrawl.dev",
    formats=["markdown", "html"],
)

print(doc.markdown)

Trusted by 40,000+ teams building with web data

OpenAI
Zapier
Nvidia
Shopify
Vercel
Replit
Anthropic
Perplexity

Start scraping today

Try it right here. Pick an endpoint, drop a URL, and see the exact payload your agent gets.

response · 200 OK · search
{
  "web": [
    { "title": "Firecrawl — web data for AI", "url": "https://firecrawl.dev" },
    { "title": "Docs · Search endpoint", "url": "https://docs.firecrawl.dev/search" }
  ]
}

50M+

pages crawled monthly

99.9%

API uptime

<1.5s

median scrape time

40k+

developers building

Capabilities

Everything between a URL and your model

Markdown out of the box

Boilerplate, nav and ads stripped. What lands in your context window is only the content.

Structured extraction

Pass a schema and get typed JSON back, extracted with an LLM across every crawled page.

Handles dynamic sites

JS rendering, clicking, scrolling and waiting — actions run before the page is captured.

Proxies & anti-bot

Rotating proxies, stealth mode and automatic retries so blocked pages stop being your problem.

Built for scale

Batch thousands of URLs with concurrency, caching and webhooks when the job finishes.

Screenshots & HTML

Ask for any combination of markdown, raw HTML, links, and full-page screenshots.

Fast, reliable, and token-efficient.
And it's open source

Run it on our cloud or host it yourself — same API, same output, no lock-in.

Fast

Median page in under 1.5s, with parallel batch jobs and smart caching.

Token-efficient

Boilerplate removal cuts payloads ~60%, so your model pays for signal only.

Reliable

Automatic retries, proxy rotation and 99.9% uptime across every endpoint.

People love building with Firecrawl

We replaced a whole scraping team's worth of glue code with two Firecrawl calls. Our agents finally get clean context.
MCMaya ChenHead of AI, Northwind
Token cost dropped 62% after switching to Firecrawl markdown. Same pages, none of the nav soup.
DRDiego RamosStaff Engineer, Loop
Crawl + webhooks handled 400k pages last month without a single babysitting incident.
PNPriya NairFounder, Ledgerly
The self-hosted option sealed it for us — same API, running in our VPC.
TWTom WhitfieldPlatform Lead, Arcadia
Structured extraction with a schema is absurdly good. It just returns the JSON we asked for.
SBSara BergmanML Engineer, Kite
Best DX I've had with a data API in years. Five minutes from key to first crawl.
JOJules OkaforCTO, Fieldnote

Pricing

Pay for pages, not for infrastructure

Free

$0

500 credits, one-time

  • Scrape, crawl & map
  • 2 concurrent browsers
  • Community support
Start free

Hobby

$16/mo

3,000 credits / month

  • 5 concurrent browsers
  • Markdown, HTML & JSON
  • Email support
Choose Hobby

Standard

Popular

$83/mo

100,000 credits / month

  • 50 concurrent browsers
  • Structured JSON extraction
  • Webhooks & batch jobs
  • Priority email support
Choose Standard

Enterprise

Custom

volume pricing & SLA

  • Unlimited concurrency
  • SSO & audit logs
  • Self-host option
  • Dedicated engineer
Talk to us

Compare every feature

FeatureFreeHobbyStandardEnterprise
Usage
Credits included500 once3,000 /mo100,000 /moCustom
Concurrent browsers2550Unlimited
Rate limit10 rpm60 rpm600 rpmCustom
Endpoints
/scrape, /map, /crawl
/search
/extract (JSON)
Agent runs
Platform
Webhooks
Activity logs retention24h7 days90 daysCustom
SSO & audit logs
SupportCommunityEmailPriorityDedicated
Get startedGet startedGet startedContact

Frequently asked questions

What exactly does Firecrawl return?+

Clean markdown, raw HTML, links, screenshots and — when you pass a schema — structured JSON. Everything is stripped of nav, ads and boilerplate so it fits in a context window.

Do I need my own proxies or browsers?+

No. Rotating proxies, stealth mode, JS rendering and retries are handled for you. You send a URL, you get content back.

How is a credit counted?+

One credit per successfully scraped page. Crawls charge per page fetched; failed requests are never billed.

Can I self-host it?+

Yes. Firecrawl is open source under AGPL-3.0 and runs in your own infrastructure with the same API surface.

Is there a free tier?+

Every account starts with 500 free credits, no credit card required.

Stay in the loop

Book a demo or get the crawler digest

Tell us where to reach you. We'll send a confirmation and either schedule a walkthrough or add you to our monthly product digest — no spam, unsubscribe anytime.

By submitting you agree to receive emails about Firecrawl. Unsubscribe anytime.

Ship your first crawl in five minutes

Grab an API key, paste a URL, and get clean markdown back. Free credits included.