Firecrawl Review (2026): What It Does, What It Costs and Who It Fits

Some links on this page are affiliate links. If you sign up through them, we may earn a commission at no extra cost to you. How we make money

Firecrawl is a web data API that turns websites into clean Markdown or structured JSON for AI tools, scripts and sales workflows. It is a strong pick for teams that want one API for scraping, crawling, web search, page monitoring and AI agents, with a free plan and paid plans from $19 a month (as of November 2026). It is a weaker fit if you need a no-code point-and-click scraper, predictable flat costs for AI extraction, refunds, or data from social platforms.

Some tool links here are affiliate links (disclosure). Nerdy Joe is a Firecrawl affiliate, and also an affiliate of Findymail, Apollo, Instantly, Smartlead and PlusVibe. If you want to follow along with the plans, open Firecrawl's pricing page.

Nerdy Joe is a certified Apollo.io Managed Services Partner and a PlusVibe Certified Silver Partner, and is owned by Agency Growth Partners (AGP). Our relationship with Firecrawl is affiliate only. We did not test Firecrawl hands-on for this review; everything below comes from Firecrawl's site, docs, GitHub, terms and integration pages as fetched November 20 and 21, 2026, plus the third-party and competitor pages listed in Sources.

Firecrawl review: the short verdict

Firecrawl is a good choice when you want web pages in a format an AI model can read, and you want scraping, crawling, search and monitoring behind one API key. The credit model is simple for basic pages (1 credit per page) and gets less predictable once you add AI formats, agents or browser sessions.

The trade-offs are in the fine print. Credits do not roll over on most plans, the terms say there are no refunds "for any reason," and Firecrawl's own pages disagree on a handful of prices and rules, which we list below.

Question Short answer (as of November 2026)
What it is An API that scrapes, crawls, maps, searches and monitors the web, with AI agents on top
Plans (monthly billing) Free $0, Hobby $19, Standard $99, Growth $399, Scale $749, Enterprise custom
Annual billing (per month) Hobby $16, Standard $83, Growth $333, Scale $599
Free plan 1,000 credits a month, no card
Open source Yes, main repo primarily AGPL-3.0; cloud has extra features
No-code options Zapier, Make, n8n, Pipedream; Instantly built its own integration
Official Clay integration None found
Best for Developers, GTM engineers and AI builders who need web data at volume
Weaker for Non-technical users who want a visual scraper, and social media data

For every plan, credit cost and add-on, see our full Firecrawl pricing breakdown.

What Firecrawl is in 2026

Websites are built for people, full of menus, scripts and cookie banners, while an AI model or a spreadsheet only needs the words, links and a few facts. Firecrawl sits in the middle: you send it a URL or a question, and it sends back the page as clean Markdown (plain text with simple formatting), raw HTML, a list of links, metadata, a screenshot, or JSON fields you define, such as company name, pricing tiers or open roles.

On top of that, Firecrawl adds web search, an AI agent, browser sessions that click and fill forms, page-change monitoring, and Alexandria, a people and company data exchange. Firecrawl says it is "Used by over 1.25M developers," a claim we cannot check.

Who Firecrawl is for

  • GTM engineers and RevOps who build enrichment and research flows in code, n8n, Zapier or Make.
  • Agencies running outbound that want live website data (pricing, careers pages, news) to research accounts and write openers.
  • AI builders and coding-agent users who want web content for chatbots or agents, or want Claude Code or Cursor to read the web through Firecrawl's MCP server.

Who should look elsewhere

  • Non-technical users who want to click on a page and save a table. A no-code tool such as Browse AI is closer to that.
  • Teams that need social platform data. A third-party test reported Instagram, YouTube and TikTok as unsupported, and we could not confirm LinkedIn support either way.
  • Buyers who need a contact database. Firecrawl reads websites; for verified emails you still want a data provider or email finder such as Apollo or Findymail, which we cover in our best B2B data providers guide.

Every Firecrawl endpoint in plain English

An "endpoint" is one job the API can do, called with one request. Here is what each one does, using Firecrawl's own descriptions from its docs; you can sign up for Firecrawl to try each one on the free plan.

Scrape

Firecrawl describes Scrape as: "Scrape any URL and get its content in markdown, HTML, or other formats." It is the workhorse, and asking for JSON fields, a question or highlights adds 4 credits, so a structured scrape costs 5 credits.

Crawl

Crawl will "Recursively gather content from entire sites." You give it a starting URL and it follows links, reading each page it finds at 1 credit per page.

The crawl limit defaults to 10,000 pages, and if the starting URL fails your includePaths filter, the docs warn "the crawl may return 0 pages."

Map

Map will "Discover all URLs on a website." It is a cheap first step: list every page a company has, then scrape only the pricing, careers or about pages you need (its cost is one of the conflicts below).

Search lets you "Search the web and get full page content from results in one call," across web, image and news results. For sales teams, this is how you find recent news or press about an account without a separate search API.

Agent

Agent is "Autonomous web data gathering powered by AI." You describe what you want, such as a company's founders, and the agent decides which searches and pages to run.

The default model is spark-2, and Firecrawl says "Spark 1 models are deprecated." You can set an effort level of low, medium or high, and a maxCredits cap that defaults to 2,500 credits.

Pricing is "dynamic": the docs say most runs cost "a few hundred credits," while an endpoint-comparison example estimates "~15 credits" for a founders lookup. A run that hits maxCredits fails and returns no data (a partial result is possible), results stay in the API for only 24 hours, and shared run links are public.

Interact

Interact lets you "Continue working with any scraped page: click, fill forms, extract dynamic content," such as pressing "load more" before reading the page.

It is billed by browser minute, and Browser Sandbox, a related product, offers "Managed browser sessions for interactive workflows."

Monitor

Monitor sends a webhook (an automatic message to another app) when a page changes. Firecrawl's lead-enrichment page gives the sales use: watch a prospect's pricing or careers page and get alerted when it moves.

Parse and Webhooks

Parse will "Turn local PDFs, DOCX, XLSX, HTML, and more into Markdown or structured JSON," which helps with files you already have, such as a conference attendee list. Webhooks provide "Async event delivery," so long crawls can notify your system when they finish.

/extract is shutting down on December 10, 2026

If you built anything on /extract, plan a move now. Firecrawl's docs say: "/extract is deprecated and will be removed on Thursday, December 10, 2026, 15:00 UTC."

Single-page extraction moves to /v2/scrape with JSON format. Multi-page or web-search extraction moves to /v2/agent, which the docs call "the successor to /extract." Zapier and Make still listed Extract actions when we checked, so review any zaps or scenarios that use them.

Output formats

Firecrawl's docs list Markdown, HTML, metadata, links, structured JSON and screenshots, and the billing docs price extra formats and add-ons such as PII redaction and a prompt-injection check (useful because pages can contain text written to manipulate AI tools).

Open source vs cloud

Firecrawl's core is open source, and its main repo had 200.6k GitHub stars and 10.6k forks (as of November 20, 2026). The main repository is "primarily licensed under the GNU Affero General Public License v3.0 (AGPL-3.0)," while "The SDKs and some UI components are licensed under the MIT License."

AGPL-3.0 is a copyleft license. In plain terms, if you modify Firecrawl and run it as a service for others, the license can require you to share your changes; ask a lawyer if that applies to you.

The cloud version "includes additional features." The repo shows the exact differences only in an image, so we cannot list them with confidence. If you plan to self-host, confirm that the endpoints you need (Agent, Monitor and Alexandria in particular) are in the self-hosted build.

For most sales teams, the cloud version is the sensible start: create a free Firecrawl account and try the core endpoints before deciding whether self-hosting is worth the work.

MCP server, SDKs and integrations

The Firecrawl MCP server

MCP (Model Context Protocol) is a standard that lets AI assistants such as Claude or Cursor call outside tools. Firecrawl's MCP server is licensed MIT and had 7.8k GitHub stars as of November 20, 2026.

The hosted URL is https://mcp.firecrawl.dev/v2/mcp, with an OAuth variant at /v2/mcp-oauth. Keyless access covers only scrape, search and parse; the full set of 29 tools (including crawl, map, agent, interact, monitor and people and company search and enrichment) needs a FIRECRAWL_API_KEY, or FIRECRAWL_API_URL if you self-host.

For a GTM operator, this lets Claude Code or a similar assistant research accounts by calling Firecrawl directly, which is the setup Firecrawl's own prospecting guide uses.

SDKs

Official SDKs cover Python, Node.js and a CLI, and the repo also lists Go, Java, Elixir, Rust, Ruby, .NET and PHP.

No-code: Zapier, Make, n8n and Pipedream

Firecrawl's integrations page lists Zapier, Make, n8n and Pipedream for workflow automation. The details differ by platform.

Platform What is available Who maintains it
Zapier Actions only, no triggers: Scrape Page, Crawl Website, Map Data, Search Data, Extract Data, Run Agent, Get Agent Status Not stated on the page we used
Make Crawl, Extract, Get Crawl Status, Get Extract, Make an API Call, Map, Scrape, Search Not stated
n8n Verified community node, "built and maintained by our partners at Firecrawl and verified by n8n" Firecrawl, verified by n8n
Pipedream Listed on Firecrawl's integrations page Not checked

No triggers on Zapier means something else, such as a new CRM record or a schedule, has to start the zap. n8n also shows a template that scores and routes new trial leads with Firecrawl, Scrapin.io, OpenAI, Slack and Instantly.

Instantly's Firecrawl integration

Instantly lists a Firecrawl integration in its marketplace, marked as built by Instantly, and it uses your own Firecrawl account. It has two triggers (Page Changed and Job Finished) and six actions (Scrape URL, Search the Web, Map Website, Crawl, Batch Scrape and Run Agent).

That matters for outbound, because a Page Changed trigger on a monitored careers or pricing page can start a workflow in your sender; our Instantly review covers the sending side, and you can see Instantly's plans if you need a sender.

Clay: no official integration found

Neither Firecrawl's integrations list nor Clay's integrations directory lists the other. Firecrawl mentions Clay only as a destination on its lead-enrichment page and as a price comparison.

Calling Firecrawl from Clay through a generic HTTP API column may be possible, but we found no official documentation for it. Test it before you plan a workflow around it; our Apollo vs Clay comparison covers where Clay fits.

Agent frameworks, builders and app platforms

Firecrawl also lists agent frameworks (LangChain, LlamaIndex, CrewAI, CAMEL-AI, Praison AI, Neuron AI, Vercel AI SDK, Mastra), visual builders (Dify, Langflow, Flowise), app builders (Replit, Lovable, Vercel Marketplace, Stripe Projects) and Composio.

Firecrawl pricing as of November 2026

All prices come from Firecrawl's pricing page as fetched November 20, 2026, in USD.

Plan Monthly billing Annual billing (per month) Credits a month Concurrency Extra credits per $5 Rollover
Free $0 $0 1,000 2 Not offered No
Hobby $19 $16 5,000 5 1,000 No
Standard $99 $83 100,000 25 2,000 No
Growth $399 $333 500,000 50 2,500 No
Scale $749 $599 1,000,000 100 5,000 One month (see conflicts)
Enterprise Custom Custom Custom Custom Custom Custom

The free plan is "No cost, no card, no hassle," and its credits refresh monthly. Search and scrape draw on one balance, so 1,000 free credits is about 1,000 basic pages or 500 searches.

What each action costs in credits

Action Credits
Scrape or crawl, basic page 1 per page
Map 1 per call (one pricing table says 1 per page)
Search 2 per 10 results, rounded up, plus pages scraped
JSON, question or highlights format +4 per page
PDF parsing +1 per PDF page
Zero Data Retention +1 per page
Enhanced proxy +0
Audio, video, prompt-injection check, PII redaction +4 each
Interact 2 per browser minute (code), 7 (prompt), one-minute minimum
Monitor 1 per page per check; 7 flat on the deterministic engine
Agent Dynamic; 5 free runs a day; maxCredits default 2,500

Firecrawl does not list costs for stealth proxies or screenshots on the pages we fetched. A scrape that returns no result is not charged, but a 403 or 404 page that is still returned costs 1 credit, and the docs say some checks (threat scan 2 credits, prompt-injection check 5 credits, lockdown miss 1 credit) can be charged even when no document comes back.

Cost per 1,000 basic pages (our arithmetic)

This is our arithmetic from list prices, assuming every credit goes to 1-credit basic pages. Your real cost depends on formats and add-ons.

Plan Monthly billing per 1,000 pages Annual billing per 1,000 pages
Hobby $3.80 $3.20
Standard $0.99 $0.83
Growth $0.80 $0.67
Scale $0.75 $0.60

Structured JSON costs 5 credits a page, so multiply those figures by 5 for AI extraction. On Hobby, 1,000 JSON pages would use the full 5,000 monthly credits.

Cost per 1,000 enriched accounts (our arithmetic)

Here is a simple account-research pattern, priced from the credit table. It is our example, not a Firecrawl benchmark.

  • One search for recent news, 10 results: 2 credits.
  • Three pages per account (home, about, careers) scraped as JSON at 5 credits each: 15 credits.
  • Total: 17 credits per account, or 17,000 credits per 1,000 accounts.

On Standard, 100,000 credits covers about 5,800 accounts a month in this pattern, which works out to about $16.83 of plan value per 1,000 accounts at the monthly price. Add a Map call per account (1 credit) and it rises to 18 credits; add Agent runs, Alexandria people data, or AI model costs and the numbers change, so test on 50 accounts first.

Extra credits, overage and rollover

Extra credits are sold in $5 packs. By our arithmetic, that is $5.00 per 1,000 credits on Hobby, $2.50 on Standard, $2.00 on Growth and $1.00 on Scale.

Pay-as-you-go is for paid plans only, charged to your card in $5 steps, with an optional monthly cap; the docs say the overage allowance "is headroom, not extra credits you keep." Bought credits "expire if you cancel your subscription," and plan credits do not roll over on Free, Hobby, Standard or Growth.

Upgrades, downgrades and refunds

"Upgrades take effect immediately," at the full new-plan price with no proration, and unused time is "not credited or refunded." Downgrades take effect at renewal.

Firecrawl's terms of service (revised October 7, 2026) say: "We do not provide refunds for any reason on our Services," except a prorated refund if Firecrawl can no longer offer them. You can cancel at any time, and you can review Firecrawl's current plans before you commit.

Rate limits

Rate limits cap how many requests you can send per minute, which matters for bulk jobs.

Endpoint (requests per minute) Free Hobby Standard Growth Scale
/scrape, /map, /search 10 100 500 5,000 10,000
/crawl, /agent 2 20 100 1,000 2,000
/interact 2 20 100 1,000 1,500

Batch scrape shares the crawl limit, and extract shares the agent limit. Enterprise limits are custom.

Where Firecrawl's own pages disagree

Firecrawl's pricing page, billing docs and guides disagree in several places. We report them as found, not picking a winner; ask Firecrawl in writing if one affects your budget.

  1. Map cost. The pricing page says "1 credit per call," but its API table says "1/ page." The billing docs say 1 per call.
  2. Monthly vs annual display. The pricing page header shows annual prices ("Prices shown are per month, billed annually," Standard at $83), while the "Pricing at a glance" section shows monthly prices (Standard at $99).
  3. Concurrency label. The plan table says "Concurrent browsers," while the plan cards say "concurrent requests."
  4. Rollover. The pricing page says "Credits roll over on Scale and Enterprise" (one month on Scale). The billing docs say only annual Scale (1 month) and annual Enterprise (2 months) roll over, and that "Unused plan credits do not roll over by default."
  5. JSON mode cost. The endpoint-comparison guide says /scrape JSON mode costs "1 credit per page," described as "Fixed, predictable." The pricing page and billing docs say JSON is +4 credits per page, or 5 in total.
  6. Maps (local business data). The lead-enrichment page FAQ cites "Maps" at 5 credits per call, which does not appear on the pricing page.

Firecrawl for GTM, lead generation and sales research

Firecrawl is not only a developer tool. It publishes several pages aimed at sales and GTM teams, and this section sticks to what those pages document; our Firecrawl for lead generation guide goes step by step.

The lead-enrichment use case

Firecrawl's lead-enrichment page pitches account research from the live web, which it says stops "returning last quarter's firmographics." It also covers decision makers through Apollo and FullEnrich, verified work emails through Alexandria, buying-signal triggers through Monitor webhooks, sanctions and PEP screening, and scheduled CRM refreshes.

It names Salesforce, HubSpot, Clay, Instantly and "your warehouse" as destinations, lists 11x, Contentful and Cargo as customer stories, and claims "10x faster prospect research." The docs version lays out a three-step flow: "Point Firecrawl at a source" (a directory, company site or domain list), extract company, contact, team and news data, then "Push enriched records into Salesforce, HubSpot, or any other tool through the API or a Zapier integration." Suggested sources include chamber of commerce listings, conference attendee lists, About pages and job postings, and you can try that flow on Firecrawl's free plan.

Fire Enrich

Fire Enrich is Firecrawl's open-source enrichment template, with "over 1,100 stars" (its license is not stated on the page we read). You upload a CSV of emails and get back enriched fields, each with the source URL it came from.

It uses Firecrawl and OpenAI; Firecrawl says the self-hosted OpenAI cost is about $0.01 to $0.05 per enrichment, and the public demo is capped at 15 rows and 5 columns. Source URLs per field let a rep check a claim before it goes in an email.

Signal monitoring

Monitor plus webhooks is Firecrawl's answer to buying signals: watch a pricing, product or careers page and get a webhook when it changes. By our arithmetic, monitoring 50 pages once a day for 30 days is 1,500 checks, or 1,500 credits at 1 credit per check, or 10,500 credits on the 7-credit deterministic engine. Our guide to signal-based cold email personalization covers how to turn those alerts into relevant first lines.

The B2B prospecting workflow

A Firecrawl blog post dated November 2, 2026, by Juan Pablo Hernandez, Growth Marketer at Firecrawl, describes a five-step agent workflow that runs in Claude Code or Codex and writes to a prospects.csv file:

  1. Define the ICP with /scrape.
  2. Build the account list with Fireround, Apollo and People Data Labs.
  3. Find buyers with Apollo people search and FullEnrich emails.
  4. Check buying signals with /search news, /scrape on careers pages and Apollo job postings.
  5. Draft the opener.

It names n8n, Zapier and Make as no-code options and gives two cautions we agree with: "Never invent a company, a person or an email address," and "Read every opener and source before you send anything."

Alexandria and "Headless Clay for agents"

Alexandria is Firecrawl's tool and data exchange for people search, firmographics and sanctions or PEP screening. Providers named on Firecrawl pages are Apollo, FullEnrich, People Data Labs, Data Legion and Oriane, with Fireround named in the blog; the Alexandria docs themselves name only Particle, so the provider list depends on which page you read.

An org admin must accept Firecrawl's Third-Party Data Terms before Alexandria providers will run.

Firecrawl's November 2, 2026 changelog post, "Headless Clay for agents," pitches enrichment flows inside ChatGPT and Claude, claims search over 250M people and 26M companies plus work emails, and claims you can "Pay 88% less than Clay at scale," based on Clay Launch list prices on that date. That is Firecrawl's comparison, not a Clay partnership, and we have not checked the math.

Where Firecrawl fits in an outbound stack

Firecrawl reads the web; it does not replace a contact database, a verifier or a sender.

Job Tool type Examples
Build the account and contact list B2B database Apollo (our Apollo review)
Research accounts and catch signals Web data API Firecrawl
Find and verify emails Email finder and verifier Findymail (our Findymail review)
Send and manage replies Cold email sender Instantly, Smartlead, PlusVibe

If you need verified emails to go with Firecrawl research, you can try Findymail, and our verified lead lists for agencies guide covers list quality. For sending, see our Smartlead review and PlusVibe review.

This is a plain summary of Firecrawl's published rules, not legal advice; check with counsel for anything involving personal data.

robots.txt

robots.txt is a file where a website states which bots may visit which pages. Firecrawl's GitHub says: "By default, Firecrawl respects robots.txt directives," and the docs say it reads robots.txt before crawling, applying rules for the user-agent FirecrawlAgent and for *.

The ignoreRobotsTxt option (default false) is Enterprise only, and you must contact Firecrawl support to enable it; the same applies to a custom robotsUserAgent. The repo adds: "It is the sole responsibility of end users to respect websites' policies when scraping."

Terms of service

Firecrawl's terms (revised October 7, 2026) bar use "for any unlawful activities or in violation of any laws, regulations, or contractual provisions" (section 5.d.4). They also bar disseminating "another person's personal information without that person's permission" (5.d.10) and use "for any purpose prohibited by applicable data privacy and security laws, including the GDPR or CCPA" (5.d.15).

The terms do not mention robots.txt, getting around access controls, or a list of sites you may not scrape. "Contractual provisions" is the clause sales teams should read twice, because a website's own terms may forbid scraping.

Third-Party Data Terms: you are the data controller

Firecrawl's Third-Party Data Terms (v2.0.0, effective September 21, 2026) cover Alexandria data. Section 5.2 says that for personal data, "you process it as an independent controller (or a 'business' under the CCPA)," and Firecrawl and the provider are "not your processors."

In plain terms, if you pull people data through Firecrawl, the controller obligations under laws like GDPR or CCPA sit with you. Section 6.2 also requires you to notify Firecrawl "within seventy-two (72) hours" of a legal demand such as a cease-and-desist.

Per-source rules sit in Source Schedules (FullEnrich, Particle, Benzinga and Fiscal.ai) whose text we could not read, and the terms index page said they were "temporarily unavailable" when we checked. Read the current version before you accept.

Social platforms

A third-party test by Scrape Creators (September 18, 2025) reported that Instagram, YouTube and TikTok returned "This website is no longer supported, please reach out to help@firecrawl.com." That post cites no Firecrawl source, and we could not confirm LinkedIn support, so treat social data as out of scope unless Firecrawl confirms otherwise.

Firecrawl ratings (as of November 20, 2026)

Source Rating Volume Date checked
G2 4.7 out of 5 1,504 reviews November 20, 2026
GitHub (firecrawl/firecrawl) n/a 200.6k stars, 10.6k forks November 20, 2026
GitHub (firecrawl-mcp-server) n/a 7.8k stars November 20, 2026

Product Hunt shows 5.0 on its "firecrawl" page, but from only 3 reviews, so we do not treat it as meaningful. GitHub stars measure developer interest, not product quality; the G2 score is the better buyer signal. (G2's description cites "more than 190,000 GitHub stars," below the 200.6k the repo showed.)

Firecrawl pros and cons

Pros

  • One API for many jobs. Scrape, crawl, map, search, agents, browser actions and monitoring share one key and one credit balance.
  • Simple basic pricing. 1 credit per page for basic scrapes and crawls, and a free plan with no card.
  • AI-ready output. Markdown and JSON are built for language models, with source URLs in Fire Enrich.
  • Open source core. The main repo is AGPL-3.0, with MIT SDKs, so you can self-host.
  • Wide integrations. An MIT-licensed MCP server, Zapier, Make, n8n, Pipedream, agent frameworks, and an Instantly-built integration with triggers.
  • Documented GTM workflows. Lead enrichment, Fire Enrich, signal monitoring and a full prospecting guide.
  • Respects robots.txt by default. Ignoring it is Enterprise only.
  • Strong reviews. 4.7 out of 5 on G2 from 1,504 reviews as of November 20, 2026.

Cons

  • No refunds "for any reason," per the terms, and upgrades are charged in full with no proration.
  • Limited rollover. Credits do not roll over on Free, Hobby, Standard or Growth, and bought credits expire if you cancel.
  • AI costs add up. JSON and other AI formats cost 5 times a basic scrape, and Agent pricing is dynamic.
  • Page conflicts. Map cost, rollover, JSON cost and concurrency labels differ between pages.
  • /extract is going away on December 10, 2026, so older builds need migrating.
  • Short agent retention. Agent results stay in the API for 24 hours, and shared run links are public.
  • Social platforms reportedly unsupported, per a 2025 third-party test.
  • Self-hosted gaps. The cloud has extra features that are not listed in text.
  • Slow free tier. 10 requests per minute on scrape and 2 on crawl or agent.

Firecrawl vs Apify, ScrapingBee, Bright Data, Browse AI, Exa, Tavily and Crawl4AI

These tools overlap with Firecrawl but price and work differently. Prices are from each vendor's pricing page as fetched November 20, 2026; for a longer list, see our Firecrawl alternatives.

Tool Model Entry price Free tier Best for
Firecrawl Monthly credits Hobby $19 (5,000 credits) 1,000 credits a month AI-ready web data, search, agents, monitoring
Apify Platform plus usage Starter $19 a month $5 a month of usage, no card Ready-made scrapers for specific sites
ScrapingBee API credits Hobby $19 (75,000 credits) 1,000 credits, no card Raw scraping API with proxies
Bright Data Per record $1.50 per 1,000 records (pay as you go) 5,000 records a month Large-scale data with proxy infrastructure
Browse AI Credits, no code Personal $48 monthly or $19 annual 50 credits a month Non-technical users
Exa Pay per request Search $7 per 1,000 $10 of credits a month AI search and retrieval
Tavily Credits Project $30 (4,000 credits) 1,000 credits a month Search for AI agents
Crawl4AI Open source (Apache-2.0) Free to self-host Open source Developers who want a free, permissive license

Credits mean different things at each vendor, so do not compare "credits per dollar" across rows.

Firecrawl vs Apify

Apify is a platform of ready-made scrapers plus compute, with Starter at $19, Scale at $199 and Business at $999 a month, each including usage equal to its price per Apify's pricing page. It is stronger when someone has already built a scraper for the exact site you need; our Firecrawl vs Apify comparison goes deeper.

Firecrawl vs ScrapingBee

ScrapingBee sells API credits from Hobby at $19 (75,000 credits) to Business at $249 (3 million), with 1,000 free credits and no card. It fits teams that mainly need pages fetched through proxies and will parse them; Firecrawl adds search, agents, monitoring and AI formats.

Firecrawl vs Bright Data

Bright Data's Web Scraper API charges $1.50 per 1,000 records pay as you go, or $499 a month for 384,000 records, with 5,000 free records a month. Its strength is scale and proxy infrastructure; Firecrawl is lighter to start and more focused on AI-ready output.

Firecrawl vs Browse AI

Browse AI is no-code: Personal is $48 a month billed monthly or $19 billed annually, with a free plan of 50 credits a month, though its credit counts differ between its monthly and annual tables. It is the better fit if nobody on your team writes code.

Firecrawl vs Exa and Tavily

Exa charges per request (search $7 per 1,000, page contents $1 per 1,000) with $10 of free credits a month. Tavily starts at $30 for 4,000 credits with 1,000 free credits a month, per its docs, since its pricing slider rendered garbled; both are search-first, while Firecrawl also crawls, clicks and monitors.

Firecrawl vs Crawl4AI

Crawl4AI is open source under Apache-2.0, a more permissive license than AGPL-3.0, with 84.6k GitHub stars as of November 20, 2026. It suits developers happy to run their own crawler; Firecrawl suits teams that want a managed service with search, agents and integrations.

Firecrawl verdict by use case

Your situation Verdict
GTM engineer building enrichment in code or n8n Strong fit; one API for scrape, search, map and monitor
Agency researching accounts for personalized outbound Good fit, paired with a data provider, verifier and sender
Team that wants signal alerts (pricing, careers changes) Good fit with Monitor; Instantly's integration has a Page Changed trigger
AI builder needing web content for a model Strong fit; Markdown and JSON output plus the MCP server
Non-technical user who wants a visual scraper Weaker fit; Browse AI is closer
Anyone relying on Instagram, YouTube or TikTok data Poor fit, per a 2025 third-party test
Buyer who needs refunds or rollover on mid-tier plans Poor fit; start on the free plan or monthly Hobby

If you are in one of the first four rows, start on Firecrawl's free plan and move to Hobby or Standard once you know your credit use.

How AGP would set up Firecrawl for outbound

This is how we would set up Firecrawl based on its documentation, not a report of work we ran in it.

  1. Start free. Use the 1,000 free credits to test scrape, map and search on 20 to 50 real accounts from your ICP.
  2. Map, then scrape. Run Map on each domain and scrape only signal pages (pricing, careers, about, news) as JSON with a short schema, since each JSON page costs 5 credits.
  3. Store source URLs. Keep the URL for every fact, as Fire Enrich does, so reps can check it.
  4. Get contacts elsewhere. Pull buyers from Apollo or Alexandria, then verify emails with a dedicated verifier before sending.
  5. Set up Monitor on your best accounts. Watch careers and pricing pages for the top tier and route changes to Slack or your sender.
  6. Cap spend. Set a pay-as-you-go cap in $5 steps and a maxCredits limit on every Agent run.
  7. Review before sending. Follow Firecrawl's own advice and read every opener and source before anything goes out.
  8. Check the plan before upgrading. Upgrades are charged in full with no proration and there are no refunds, so upgrade early in a billing cycle.
  9. Migrate off /extract. Move any workflow that uses it before December 10, 2026.

When your test looks good, pick a Firecrawl plan that covers about one and a half times your measured monthly credits, so you leave room before overage starts.

How we compiled this review

We used Firecrawl's pricing page, docs, blog, changelog, GitHub repositories, terms of service and Third-Party Data Terms as fetched November 20 and 21, 2026, plus the n8n, Zapier, Make and Instantly integration pages, G2, Product Hunt, a Scrape Creators test and each competitor's pricing page.

We did not use Firecrawl hands-on, ran no accuracy or speed tests, and have no screenshots. Where Firecrawl's pages disagree, we say so; cost examples are labeled as our arithmetic. Several Firecrawl pages contained text addressed to AI agents, which we ignored.

Rather not build this yourself?

A scraping API is one piece. Building the account list, researching signals, verifying emails, writing copy and running inboxes is a weekly job.

Nerdy Joe is owned by Agency Growth Partners (AGP), and AGP builds lists, runs outbound and books meetings for agencies. If you would rather hand that off, book a call with David, or see how the pieces fit in our agency growth machine overview.

Verdict

Firecrawl is a capable, well-rated web data API that covers scraping, crawling, mapping, search, AI agents, browser actions and page monitoring under one credit balance. For GTM teams it is most useful as the research and signal layer between a contact database and a sender, and its docs and MCP server make that easy to wire into Claude Code, n8n, Zapier or Instantly.

The weak spots are commercial and practical. There are no refunds, rollover is limited, AI formats cost 5 times a basic page, Agent costs are dynamic, /extract ends December 10, 2026, and several pages disagree on costs and rules.

If you have someone who can build a simple workflow, try Firecrawl free, measure your credits on a real account list, and buy the smallest plan that fits. If you want the full cost breakdown first, read our Firecrawl pricing guide, or compare options in Firecrawl alternatives.

Disclosure: My relationship with Instantly is that of an independent party. I do not hold any position, nor do I have the power to form binding agreements or act on behalf of Instantly. My earnings are derived from referral compensations provided by Instantly. The views I share are solely mine and should not be seen as official statements, promises, or representations of Instantly, or any related entities or personnel.

Sources

Frequently asked questions

Is Firecrawl free?

Firecrawl has a free plan with 1,000 credits a month (about 1,000 basic pages or 500 searches), 2 concurrent requests and no card required, as of November 2026. Paid plans start at $19 a month on Hobby, or $16 a month billed annually.

What is Firecrawl used for?

Firecrawl turns web pages into clean text, Markdown or structured JSON that AI tools and scripts can use. Teams use it to scrape single pages, crawl whole sites, search the web, run AI agents that gather data, and monitor pages for changes.

Is Firecrawl open source?

Yes, partly. The main Firecrawl repository is primarily licensed under AGPL-3.0, with the SDKs and some UI components under MIT. Firecrawl says its cloud version includes additional features that the self-hosted version does not.

Does Firecrawl respect robots.txt?

Yes, by default. Firecrawl's docs say it reads robots.txt before it crawls a site, and the option to ignore robots.txt is Enterprise only and must be enabled by Firecrawl support.

Is the Firecrawl /extract endpoint being removed?

Yes. Firecrawl's docs say /extract is deprecated and will be removed on Thursday, December 10, 2026, at 15:00 UTC. Single-page work moves to /v2/scrape with JSON format, and multi-page work moves to /v2/agent.

Does Firecrawl have a Clay integration?

We found no official one as of November 2026. Clay's integrations directory does not list Firecrawl, and Firecrawl's integrations page does not list Clay, although Firecrawl names Clay as a destination on its lead-enrichment page.