Your AI agent needs the web, and Firecrawl turns any page into clean Markdown in one API call. So what will next month cost?
Get Firecrawl pricing wrong and you’ll hit a rate limit mid-crawl. Or you’ll pay five times more per credit than you need to.
We used the Firecrawl API while researching this piece. Every figure here comes from its rate card, which changed on September 4, 2026.
The cheapest plan for you comes down to one number most buyers never work out.
Firecrawl is free for 1,000 credits a month, and one credit scrapes one page. Paid plans cost $19 (Hobby), $99 (Standard), $399 (Growth) or $749 (Scale) a month. Billed yearly, those drop to $16, $83, $333 and $599.
Every endpoint works on every plan, even Free. What changes between plans is volume, speed and the price of each credit.
If you sign up through our links, we may earn a commission. Your price stays exactly the same, and our affiliate disclosure explains how it works.
What does Firecrawl pricing look like in real dollars?
Here’s the full rate card. Monthly prices come first, since that’s what you pay without a yearly commitment.
Firecrawl plans and prices
Monthly list prices, effective September 4, 2026
- Included: 1,000 credits every month
- Included: 2 concurrent browsers
- Included: Every endpoint included
- Not included: Pay-as-you-go top-ups
- Included: 5,000 credits a month
- Included: 5 concurrent browsers
- Included: Top-ups: 1,000 credits per $5
- Not included: Credit rollover
- Included: 100,000 credits a month
- Included: 25 concurrent browsers
- Included: Top-ups: 2,000 credits per $5
- Not included: Credit rollover
Growth costs $399 a month for 500,000 credits. Scale costs $749 for 1 million credits, with one month of rollover. Yearly billing saves 16.7%, or 20% on Scale.
We may earn a commission when you buy through links on this page. This does not affect our reviews.
| Plan | Monthly | Billed yearly | Credits a month | Concurrent browsers |
|---|---|---|---|---|
| Free | $0 | $0 | 1,000 | 2 |
| Hobby | $19 | $16/mo ($190/yr) | 5,000 | 5 |
| Standard | $99 | $83/mo ($990/yr) | 100,000 | 25 |
| Growth | $399 | $333/mo ($3,990/yr) | 500,000 | 50 |
| Scale | $749 | $599/mo ($7,190/yr) | 1,000,000 | 100 |
| Enterprise | Custom | Custom | Custom | Custom |
The jump from Hobby to Standard looks steep at $80 more a month. For that, you get 20 times the credits and five times the concurrency.
Growth adds a shared Slack channel and much higher rate limits. Scale is the only self-serve plan that rolls unused credits into the next month. It can also pay by invoice instead of card.
How do Firecrawl credits work?
One credit buys one basic page, whether you scrape it, crawl it or map a site. Everything else is priced as a multiple of that.
All endpoints draw on one shared balance. Searches, scrapes and browser minutes all come out of the same pot, at the same rate on every plan.
| What you do | Credits | Notes |
|---|---|---|
| Scrape a page | 1 | Markdown, HTML or links |
| Crawl a site | 1 per page | Each page crawled |
| Map a site | 1 per call | Returns the URL list |
| Search the web | 2 per 10 results | Scraping the results costs extra |
| JSON, question or highlights format | +4 per page | Charged per format |
| PII redaction | +4 per page | Strips personal data |
| Audio or video extraction | +4 per page | |
| PDF parsing | 1 per PDF page | |
| Enhanced anti-bot proxy | +0 | Same 1 credit as basic |
| Interact (browser session) | 2 per minute | Clicks and typing in a live browser |
| Monitor (change tracking) | 1 per page per check | 7 with deterministic extraction |
| Agent (preview) | Dynamic | First 5 runs a day free |
The extras stack. Ask for JSON and a question answer on the same page, and it costs 9 credits instead of 1.
Our own usage matched the table. While researching this article through Firecrawl’s hosted MCP server, a plain Markdown scrape reported 1 credit. The same kind of page with a question format reported 5, and a five-result search reported 2.
One pleasant surprise: the enhanced proxy for anti-bot sites costs nothing extra. Firecrawl bills it at the same 1 credit, and an automatic retry through it isn’t charged twice. Run your own scraper and you’d pay for those proxies separately, as our guide to proxies for web scraping explains.
What does one credit really cost on each plan?
A credit costs $3.80 per 1,000 on Hobby but only $0.99 per 1,000 on Standard, billed monthly. Top-ups cost more than either.
| Plan | Per 1,000 credits (monthly) | Per 1,000 credits (yearly) | Top-up per 1,000 |
|---|---|---|---|
| Hobby | $3.80 | $3.17 | $5.00 |
| Standard | $0.99 | $0.83 | $2.50 |
| Growth | $0.80 | $0.67 | $2.00 |
| Scale | $0.75 | $0.60 | $1.00 |
Remember that number most buyers never work out? It’s 21,000 credits a month.
Hobby includes 5,000 credits for $19, and each extra 1,000 costs $5. At 21,000 credits, that’s $19 plus $80 in top-ups, exactly Standard’s $99. Past that point, Hobby only gets more expensive.
Billed yearly, the crossover comes a little sooner, at roughly 18,000 credits. Either way, a Standard credit costs about a quarter of a Hobby one.
What do real scraping jobs cost?
Rate cards are abstract. Here’s what six common jobs cost at list prices, billed monthly, on the cheapest setup that covers them.
| Job | Credits a month | Cheapest setup | Monthly cost |
|---|---|---|---|
| Scrape 3,000 blog posts as Markdown | 3,000 | Hobby | $19 |
| Crawl a 2,000-page docs site weekly | About 8,700 | Hobby + 4 top-ups | $39 |
| Monitor 500 pages daily for changes | 15,000 | Hobby + 10 top-ups | $69 |
| 5,000 searches, scraping the top 5 results | 35,000 | Standard | $99 |
| Extract JSON from 10,000 product pages | 50,000 | Standard | $99 |
| Scrape 1 million pages | 1,000,000 | Scale | $749 |
See the pattern? Hobby with top-ups wins for small, steady jobs. Once structured output or search-plus-scrape enters the picture, Standard pays for itself fast.
The search example surprises people. Each search costs 2 credits, but scraping its five results adds 5 more.
Need only result links and snippets? A dedicated SERP API can be cheaper, and our list of the best SERP APIs compares them.
Here’s the catch: what the plan price doesn’t cover
Firecrawl’s pricing page is unusually clear. Still, five details can make a bill bigger than you planned:
- Pay-as-you-go has no cap by default. Top-ups arrive in $5 steps whenever you run out. Leave the monthly limit blank and there’s no ceiling, so set one on day one.
- Plan credits don’t roll over. Unused Hobby, Standard and Growth credits vanish at the end of each period. Only Scale carries them over, and only for one month.
- There are no refunds. Firecrawl’s terms of service say it doesn’t refund for any reason. You can cancel anytime, and charges stop at the end of the billing period.
- Extras multiply fast. JSON, question and highlights formats, PII redaction, and audio or video extraction each add 4 credits per page. A crawl with JSON turned on costs five times a plain one.
- The Agent endpoint is hard to budget. After 5 free runs a day, it bills credits by the tokens the agent uses. Similar tasks can cost different amounts.
How to keep your Firecrawl bill predictable
Five habits that stop credit surprises
-
Step 1 of 5
Prototype on free credits
Build and test on the 1,000 free credits each month. Keyless calls work too, within daily limits per IP address.
-
Step 2 of 5
Measure one real job
Run your actual workload once and note the credits it reports. Multiply by how often it runs each month.
-
Step 3 of 5
Request only the formats you need
Stick to Markdown unless you need structured data. Each extra format adds 4 credits per page.
-
Step 4 of 5
Cap pay-as-you-go
In Billing, set a monthly limit in multiples of $5. Enter 0 to switch top-ups off completely.
-
Step 5 of 5
Upgrade at the crossover
Buying more than 16 Hobby top-ups a month? Move to Standard, where each credit costs about a fifth of a top-up credit.
Does paying more buy you a faster API?
Yes. Higher plans raise both concurrency and requests per minute, and the gap between tiers is big.
| Plan | Concurrent browsers | Scrape, map, search (per min) | Crawl, agent (per min) |
|---|---|---|---|
| Free | 2 | 10 | 2 |
| Hobby | 5 | 100 | 20 |
| Standard | 25 | 500 | 100 |
| Growth | 50 | 5,000 | 1,000 |
| Scale | 100 | 10,000 | 2,000 |
Go over a limit and the API answers with a 429 error, so build retries into your code. Batch scrapes share the crawl limit, and extract shares the agent limit.
For agents that fire searches in parallel, concurrency matters more than credits. A Free key’s 10 searches a minute runs out in seconds.
Is there a free way to use Firecrawl?
Yes, four of them, and each has a limit worth knowing.
- The Free plan. You get 1,000 credits every month with no card. That’s about 1,000 pages or 500 searches.
- Keyless calls. The API works without a key at all, capped by daily request and credit limits per IP address.
- Five free Agent runs a day. The Agent endpoint is in preview, and your first 5 runs each day cost nothing.
- Self-hosting. Firecrawl is open source under the AGPL-3.0 license, so you can run it on your own servers.
Self-hosting isn’t free in practice, though. You’ll run PostgreSQL, Playwright and a message queue yourself, and the default stack skips screenshots and page actions. Agent, browser interaction and some specialized formats stay cloud-only.
Prefer point-and-click to code? Our Thunderbit review covers a no-code AI scraper with its own free tier.
Can you get a Firecrawl discount?
The only public discount is yearly billing. It takes 16.7% off Hobby, Standard and Growth, and 20% off Scale.
We found no public coupon codes or startup program on Firecrawl’s site. Bigger plans do get cheaper top-ups, though: $5 buys 1,000 credits on Hobby but 5,000 on Scale.
Enterprise buyers can negotiate bulk discounts, according to the pricing page. That’s worth a sales call once you’re spending thousands a month.
Firecrawl vs ScraperAPI vs Apify: who’s cheaper for your job?
For clean Markdown an AI model can read, Firecrawl is the simplest buy. ScraperAPI is cheaper per raw page, and Apify wins when a ready-made scraper already exists for your target.
| Feature | |||
|---|---|---|---|
| Free option | 1,000 credits a month | 7-day trial, 5,000 credits | $5 of usage a month |
| Entry plan includes | 5,000 credits | 100,000 API credits | $19 of usage |
| 1,000 simple pages (entry plan) | $3.80 | $0.49 | About $0.20 to $5 |
| Billed by | Credits per page or search | Credits per request | Compute units and Actor fees |
We may earn a commission when you buy through links on this page. This does not affect our reviews.
Pick Firecrawl for LLM and agent pipelines, because it returns Markdown or JSON with no parsing code.
Pick ScraperAPI if you need raw HTML at high volume, because 100,000 basic requests cost $49 a month. Protected targets cost more, though: a Google request uses 25 credits. Our ScraperAPI review has the details.
Pick Apify if a scraper already exists for your target, because Store Actors charge per result. Google Maps places cost $3 per 1,000 on Starter. General crawling runs about $0.20 to $5 per 1,000 pages, by Apify’s estimate.
Want the wider field? Our roundup of web scraping tools covers more options, from no-code apps to raw proxy APIs.
Which Firecrawl plan should you pick?
Pick by monthly credits, not by features, because every plan gets every endpoint.
- Free for testing the API or building a demo, because 1,000 credits a month can prove an idea.
- Hobby for side projects under about 21,000 credits a month, because small top-ups beat a $99 plan below that line.
- Standard for agents, RAG pipelines and anything using JSON output, because its credits cost a quarter of Hobby’s.
- Growth or Scale past 500,000 credits or when rate limits bite, because both the credit price and the limits improve.
- Enterprise if you need SSO, zero data retention or a signed SLA, because no self-serve plan offers them.
How we checked these numbers (and what we didn’t)
We used Firecrawl’s scrape and search endpoints through its hosted MCP server while researching this article. We watched the credit counts it reported.
Every price comes from Firecrawl’s pricing page, its billing, rate-limit and self-hosting docs, and its terms of service. All of it was checked on September 28, 2026. Competitor prices come from ScraperAPI’s and Apify’s own pricing pages on the same day.
What we didn’t do: run a paid plan for a full month or benchmark speed per plan. Prices change, so check the official page before you commit a budget.
Frequently asked questions
Does Firecrawl charge for failed requests?
Only when a page comes back: a scrape that returns no result costs nothing. A page that answers with an error such as 403 or 404 still comes back, so it costs 1 credit. If a charge looks wrong, Firecrawl asks you to email [email protected].
Do Firecrawl credits roll over to the next month?
Plan credits don’t, except on Scale, which carries unused credits for one month, and Enterprise, which sets its own terms. Credits you buy through pay-as-you-go work differently. They stay on your account until you use them, but they expire if you cancel your subscription.
What payment methods does Firecrawl accept?
Firecrawl bills through Stripe, which takes most major credit and debit cards plus PayPal. Scale and Enterprise customers can switch to invoiced billing against a signed order form instead of a card. Pay-as-you-go top-ups always charge the card on file in $5 increments.
Does Firecrawl pricing include taxes?
No, prices are in US dollars, and local tax such as VAT is added automatically based on your location. A valid tax ID removes that tax in some regions, but not everywhere, so it won’t always lower the amount. Firecrawl prints your tax ID on the invoice either way.
Is Firecrawl secure enough for company data?
Every plan, including Free, is covered by Firecrawl’s SOC 2 Type II audit, multi-factor sign-in and signed webhooks. Standard and above get a countersigned data processing agreement. Zero data retention, SSO and API key IP restrictions are Enterprise-only, and PII redaction costs 4 extra credits per page.
Can an AI agent sign up for Firecrawl by itself?
Yes, Firecrawl publishes an onboarding skill file for agents at firecrawl.dev/agent-onboarding/SKILL.md. Agents whose platform supports WorkOS ID-JAG tokens can sign in directly. Without a key, an agent can still search, scrape and interact on the keyless tier, within daily per-IP limits.
How is the Firecrawl Agent endpoint priced?
Agent is in preview, with 5 free runs every day. After that, it bills credits based on the tokens the agent uses, so two similar tasks can cost different amounts. Extract requests share the agent rate limit, which is 2 a minute on Free and 20 on Hobby.
Does Firecrawl charge per seat or per user?
No, every plan bills by credits rather than seats, so adding teammates to your account doesn’t raise the price. Growth and higher plans can set up multi-team organizations through Firecrawl support. Single sign-on and SCIM user provisioning are reserved for Enterprise customers.
The bottom line on Firecrawl pricing
Firecrawl pricing is simple on paper: 1,000 free credits, then $19, $99, $399 or $749 a month.
Start free and measure your usage credits per job. Move to Standard once you pass about 21,000 credits a month, because its credits cost a quarter of Hobby’s.
One warning: set a pay-as-you-go limit before you launch anything, since there’s no cap by default and no refunds.
Your next step: grab a free Firecrawl API key, run one real job and note the credits it reports.
