Firecrawl pricing

About

Firecrawl is a web-scraping and data-extraction API built for AI agents and developers. It turns any URL — or an entire site — into clean, LLM-ready markdown or structured JSON through a small set of endpoints: Scrape (single page), Crawl (whole site), Map (URL discovery), Search (web search to results), Interact (browser automation), and Monitor (change detection). The pitch is that web data is the hard part of building AI applications, and Firecrawl removes the boilerplate of headless browsers, proxies, and parsing so a developer can go from prompt to structured data in a few lines of SDK code.

Founded in 2022 and based in San Francisco, Firecrawl was built by Caleb Peffer (CEO), Eric Ciarla (CMO), and Nicolas Silberstein Camara (CTO). It is a Y Combinator company that has raised roughly $16.2M, with Nexus Venture Partners leading its Series A alongside YC and Zapier, plus angel backing from founders including Shopify’s Tobias Lütke and Postman’s Abhinav Asthana. The product has grown bottom-up through open source: Firecrawl reports 500K+ signed-up developers and 100K+ GitHub stars, making its repo one of the most-starred on GitHub.

Competitively, Firecrawl sits in the “web data for AI” category against scraping incumbents like Apify and ScrapingBee and newer agent-data tools. Its differentiation is developer ergonomics (clean markdown out of the box, first-class SDKs and an MCP server) and a credit-based pricing model that is simple to reason about — one credit per page — rather than per-proxy or per-compute billing. Customers cited on its enterprise page include engineering teams at Zapier, Botpress, and Replit.

Pricing summary : How Firecrawl’s credit-pool subscription works

Firecrawl uses a credit-based billing model layered on monthly subscriptions: every plan buys a fixed pool of credits, and each API request burns credits — most commonly 1 credit per scraped page. There are no seats and no pure pay-as-you-go; instead the model is a hybrid of fixed monthly commitment plus optional auto-recharge usage when you exhaust the pool.

• Plan fee (the commitment): Free $0 (1,000 credits), Hobby $16/mo (5,000), Standard $83/mo (100,000), Growth $333/mo (500,000), Scale $599/mo (1,000,000), Enterprise custom. These $/mo figures are the page’s default annual-billed prices (“/monthly, billed yearly”); month-to-month rates are higher.
• Credits (the usage unit): Scrape, Crawl, Map, and Monitor cost 1 credit per page; Search costs 2 credits per 10 results; Interact costs 2 credits per browser minute; Agent is in preview with 5 free daily runs and dynamic pricing.
• Throughput (bundled with the tier): concurrent browsers scale from 2 (Free) to 150 (Scale), and per-endpoint API rate limits scale with the plan.
• Overflow (auto-recharge): when the balance drops below a threshold, auto-recharge packs buy more credits automatically, with volume discounts and 12-month validity.

What makes this different: Firecrawl collapses a messy infrastructure cost (proxies, headless browsers, parsing) into a single legible unit — the page-credit — so a developer can estimate a bill by counting pages, while the vendor still captures upside from heavy crawlers through auto-recharge rather than seats.

Pricing by product

The core product is one API billed in credits. The pricing page exposes a credit selector that lets Hobby, Standard, and Growth buyers nudge their monthly allotment up (and the price with it); the tables below capture both the base allotment and the adjustable steps read from the rendered USD pricing page. The prices below are the page’s default annual-billed rates (each paid tier shows “/monthly, billed yearly” with a “Save $X” callout); choosing month-to-month billing raises them.

Firecrawl API (self-serve plans)

Firecrawl API (scale & enterprise plans)

API credit costs (per request)

Auto-recharge (usage overflow)

Firecrawl has no pure pay-as-you-go plan. Instead, auto-recharge packs sit on top of any paid plan: when your credit balance drops below a threshold, a pack is purchased automatically so jobs never stall. Larger packs carry volume discounts, and auto-recharged credits are valid for 12 months (and do roll over, unlike base monthly credits).

Sales motions across products: PLG / self-serve for Free, Hobby, Standard, Growth, and Scale (sign up and subscribe online); sales-led for Enterprise (custom credits, concurrency, security, and SLA).

Hidden costs : When page volume and concurrency push you up a tier

The headline price hides two things: credits are consumed per page (so a single Crawl of a large site can burn thousands of credits fast), and the tier also gates concurrency, so throughput needs — not just credit needs — can force an upgrade.

A team running nightly crawls of mid-size sites:

A higher-volume data-pipeline team:

The lesson: because 1 credit = 1 page, a single broad Crawl can consume an entire Hobby or Standard pool in one run, so the real question is monthly page volume, not the sticker price — and auto-recharge can quietly add to the bill before you notice you’ve outgrown the tier.

Want to estimate your own Firecrawl bill? Use the Firecrawl pricing calculator to model your monthly cost based on pages scraped, endpoint mix, and credit allotment.

Pricing evolution : From Mendable credit packs to the seat-free tier ladder

Cadence

Tracked range: 2024-04 – 2026-06 (Wayback 2024-04-25 through the live page). Quarters not listed (2024 Q3, 2025 Q1–Q2, 2025 Q4–2026 Q1) showed the same headline tier prices across snapshots — Hobby $16, Standard $83, Growth $333 held stable; the changes were to included credit pools and concurrency, not sticker price.

Notable changes

• 2024-04 — Earliest archived /pricing (then “A product by Mendable.ai”): three credit packs only — Starter $50/mo (50k credits), Standard $375/mo (500k), Scale $1,250/mo (2.5M), priced per-1k-credit with no free tier (Wayback 2024-04-25).
• 2024-06 — Reprice to the Free $0 / Hobby $16 / Standard $83 / Growth $333 ladder under the Firecrawl brand, plus a Scrape Credits table (Scrape/Crawl/Search 1 credit/page; LLM extraction 50 credits) (Wayback 2024-06-27).
• 2024-11 — Annual billing, per-plan seats, and Auto-Recharge / Credit Pack add-ons added, moving overflow from manual bigger-pack buys to automatic top-ups (Wayback 2024-11-05).
• 2025-07 — Standard / Extract pricing tabs split out /extract billing; throughput reframed as concurrent browsers (Wayback 2025-07-07).
• 2025-08-19 — Firecrawl announced a $14.5M Series A led by Nexus Venture Partners (total raised ≈$16.2M) alongside shipping /v2 (semantic crawling, faster cached scraping) — the platform expansion that precedes the larger free/Hobby pools and the Scale tier on the current page (firecrawl.dev/blog, v2 + Series A announcement).

The 2024 reprice in detail

The single most consequential pricing event in Firecrawl’s history happened between April and June 2024, when the product was rebranded out from under Mendable.ai. The April page sold raw scraping capacity as three credit packs — $50, $375, and $1,250/month — with prices quoted per 1,000 credits and no free tier at all. Two months later the page was rebuilt around a freemium tier ladder: a $0 plan (500 credits, no card), then Hobby $16, Standard $83 (“Most Popular”), and Growth $333. The entry price to start dropped from $50 to $0, and the named tiers ($16 / $83 / $333) have held their sticker price through every snapshot since — what changed afterward was the credit pool behind each tier, not the headline number. That reprice is the foundation of the bottom-up, 500K-developer growth story: it converted a capacity-pack vendor into a freemium developer tool overnight.

Community reaction to pricing has been mixed and persistent rather than a single trust event: no Firecrawl-pricing thread has cleared Hacker News’s 50-point bar (the launch repo post and self-hosting fork Firecrawl-Simple drew 35 points, 7 comments, 2024-11), but a recurring critique is that the credit multiplier on AI extraction (historically 50 credits/page for LLM extraction, later a separate /extract billing track) makes effective per-page cost far higher than the “1 credit = 1 page” headline. A March 2026 HN post, “I replaced Firecrawl with 2,700 lines of Elixir,” captured the cost-at-scale frustration. Firecrawl’s answer has been the 2024-11 auto-recharge add-ons and, on the current page, adjustable allotments — both of which smooth spend rather than lower the unit rate.

What’s unique : One credit per page, no seats, adjustable allotments

One legible unit. Firecrawl prices the messiest part of web data — proxies, browsers, parsing — as a single page-credit. A developer can forecast spend by counting pages, which is far simpler than per-proxy or per-compute billing, and it sidesteps the bill-shock problem that plagues unpredictable AI costs.

No seats at all. Unlike most developer tools, Firecrawl never charges per user. A solo developer and a 50-person team pay the same as long as their page volume and concurrency needs match, which makes it cheap to adopt across a whole org.

Adjustable credit allotments. The pricing page lets Hobby, Standard, and Growth buyers dial their monthly credit pool up in steps (e.g. Standard from 100k to 130k to 160k credits), raising the price smoothly rather than forcing a jump to the next named tier.

Auto-recharge instead of pay-as-you-go. Rather than a metered PAYG plan, Firecrawl keeps the subscription clean and absorbs spikes through auto-recharge packs that buy credits automatically — a deliberate choice to keep the base bill predictable.

Throughput is bundled with the tier. Each plan raises concurrent browsers and API rate limits, so upgrading buys speed as well as volume — an under-appreciated second axis of the pricing.

Strengths & weaknesses

Billing UX : Credit selector, auto-recharge, and concurrency controls

• Credit selector on the pricing page — Hobby/Standard/Growth buyers can step their monthly credit allotment up (5k → 6.5k → 8k; 100k → 130k → 160k; 500k → 650k), with the price updating live.
• Currency selector — USD / EUR / GBP / INR / others; the page geo-detects a default currency and the buyer can switch it explicitly.
• Auto-recharge packs — automatic credit top-ups when the balance drops below a threshold, with volume discounts, a low-balance detection step, and 12-month credit validity.
• Concurrent-browser limits and Queue Status endpoint — each plan sets a concurrency ceiling; jobs beyond it queue (up to 48h), and a Queue Status endpoint reports availability before sending work.
• API rate limits per plan — per-endpoint requests-per-minute caps (e.g. /scrape, /crawl, /search) returning HTTP 429 when exceeded, shared across all API keys on a team.
• maxCredits parameter — caps spend on dynamic-priced Agent runs so a single query can’t overspend.
• Billing dashboard — basic dashboard analytics on all tiers, advanced analytics on Scale, plus a dedicated Slack channel for Scale and SLA support for Enterprise.

Strategic wins : Why the page-credit and free tier drove bottom-up growth

1. The page-credit is the right value metric

Firecrawl chose a unit buyers already think in — pages — instead of compute or proxies. That legibility is a core reason the product spread bottom-up; it maps cleanly to the usage-based pricing ideal of charging for the thing the customer values.

2. A genuinely usable free tier seeded 500K+ developers

1,000 credits with no card is enough to build a real prototype, which fueled open-source adoption and 100K+ GitHub stars. This mirrors the freemium playbook of landing the developer before monetizing capacity — the same usage-based pricing for SaaS and AI motion that lets the free tier double as distribution.

3. No seats removed the adoption tax

By never charging per user, Firecrawl made it trivial for a whole team to standardize on it, sidestepping the procurement friction that per-seat developer tools create — a recurring theme in our coverage of the value-metric problem in AI pricing.

Areas to improve : PAYG, rollover, and currency clarity

1. Offer a true pay-as-you-go option

Spiky and one-off workloads don’t fit monthly pools well. A metered PAYG tier (even at a premium per-page rate) would capture buyers who currently overshoot or churn, complementing the existing credit model.

2. Let base credits roll over partially

Non-rolling monthly credits punish uneven usage. A partial rollover (e.g. 1 month of carryover) would reduce waste anxiety and align with the billing-cycle fairness buyers increasingly expect.

3. Make USD pricing unambiguous

Geo-detected currency means some visitors never see the USD numbers that anchor every comparison. A clearer default (or a prominent “prices in USD” affordance) would reduce confusion for the global developer audience.

4. Surface the true cost of AI extraction up front

The most persistent community critique is that the “1 credit = 1 page” headline understates real cost once AI extraction is involved — LLM extraction historically burned 50 credits per page, and the 2025 page split /extract into its own billing track. A developer who signs up for structured extraction can find their effective credit pool is many times smaller than the sticker number. Putting an “extraction costs N× a plain scrape” example directly on the pricing page — rather than in docs — would address the recurring “egregiously expensive at scale” reaction and reduce bill-shock churn.

Key takeaways

1. Pick a unit buyers already count. Firecrawl’s page-credit works because developers think in pages, not proxies or compute — choose the value metric your customer can forecast in their head.
2. A real free tier is a growth engine, not a cost. 1,000 no-card credits seeded 500K+ developers; generosity at the bottom funded the funnel.
3. Dropping seats can be a strategy. Removing per-user fees made org-wide adoption frictionless and differentiated Firecrawl from seat-based dev tools.
4. Bundle throughput with volume. Tiering concurrency and rate limits alongside credits gives a second reason to upgrade beyond running out of credits.
5. Auto-recharge can replace PAYG — at a cost. It keeps the base bill clean but risks silent overspend, so pair it with hard caps and clear alerts.

UBP implications

1. Credit pools are the new commitment. Firecrawl shows how a monthly credit allotment functions as a soft commit-with-overage without a formal contract, smoothing the path from self-serve to enterprise.
2. Per-endpoint credit weighting is invisible price discrimination. Charging 2 credits for Search vs 1 for Scrape lets a vendor price by cost-to-serve while still advertising “1 credit per page” — a pattern other usage-based vendors can borrow.
3. Throughput-as-a-tier-axis is underused. Selling concurrency and rate limits alongside the consumption unit turns infrastructure constraints into a monetizable upgrade path.

Sources

• Firecrawl pricing page (accessed 2026-06-02)
• Firecrawl rate limits & billing docs (accessed 2026-06-02)
• Firecrawl Search endpoint docs (credit costs) (accessed 2026-06-02)
• Firecrawl Scrape endpoint docs (credit costs) (accessed 2026-06-02)
• Firecrawl Enterprise page (accessed 2026-06-02)
• Firecrawl Agent / Extract page (accessed 2026-06-02)
• Firecrawl about page (accessed 2026-06-02)
• UsagePricing blueprint corpus (accessed 2026-06-02)

Bottom line

Firecrawl turns the gnarliest part of building AI applications — getting clean web data — into a single legible unit, the page-credit, and prices it through seat-free credit subscriptions that scale from a free 1,000-credit tier to custom enterprise. It’s one of the cleanest examples in the blueprint corpus of choosing a value metric the customer already counts.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Larger free/Hobby pools, Scale tier, adjustable allotments

Jun 2026

Current live page: Free raised to 1,000 credits, Hobby to 5,000 ($16, adjustable to 6,500/8,000), Standard 100,000 ($83, adjustable to 130,000/160,000), Growth 500,000 ($333, adjustable to 650,000), a new Scale tier at $599/mo (1,000,000 credits, 150 concurrent browsers), and custom Enterprise. API credit costs: Scrape/Crawl/Map/Monitor 1 per page, Search 2 per 10 results, Interact 2 per browser minute.

Standard / Extract pricing split; concurrency reframed

Jul 2025

Wayback 2025-07-07 split the pricing page into 'Standard' (most API endpoints) and 'Extract' (/extract API) tabs, separating AI-extraction billing from page credits. Throughput was reframed from '/crawl jobs per min' to 'concurrent browsers' (Free 2, Hobby 5, Standard 50). Free still 500 credits, Hobby still 3,000 at this snapshot.

Annual billing, per-plan seats, and auto-recharge add-ons

Nov 2024

Wayback 2024-11-05 added annual billing (2 months free — e.g. Hobby $190/yr, Standard $990/yr, Growth $3,990/yr), per-plan seat counts (Standard 2 seats, Growth 4 seats), and Add-ons: Auto-Recharge Credits + a one-off Credit Pack, plus Gift Credits. Overflow shifted from 'buy a bigger pack' to automatic top-ups.

Reprice to the $16 / $83 / $333 tier ladder + free plan

Jun 2024

By Wayback 2024-06-27 the model was rebuilt under the Firecrawl brand: Free $0 (500 credits), Hobby $16/mo (3,000), Standard $83/mo (100,000, 'Most Popular'), Growth $333/mo (500,000), and a custom Enterprise plan. A Scrape Credits table appeared (Scrape/Crawl/Search = 1 credit/page; Scrape+LLM extraction = 50 credits/page). This is the tier skeleton still in use today.

Mendable-era credit packs (no free tier)

Apr 2024

Earliest archived /pricing (Wayback 2024-04-25, then 'A product by Mendable.ai'): three credit-pack plans only — Starter $50/mo (50k credits, $1.00/1k), Standard $375/mo (500k credits, $0.75/1k), Scale $1,250/mo (2.5M credits, $0.50/1k). No free plan; pricing framed per-1k-credit, not per-tier.