# Apify vs Firecrawl: per-1,000 pricing tested, plus a third column

> Apify charges 0.045 to 0.45 USD per 1,000 web results and Firecrawl 0.12 to 0.64; our live calls ran 0.069, with crawl and failures tested.

- Canonical page: [https://everyinfra.com/en/blog/apify-vs-firecrawl](https://everyinfra.com/en/blog/apify-vs-firecrawl?utm_medium=ai_doc&utm_campaign=page_md)
- Published: 2026-11-17T02:05:00+08:00 · Fact-checked: 2026-11-17T01:50:00+08:00

Take the same 1,000 web search results. Apify's Google Search Results Scraper works out to about 0.045 to 0.45 USD at public event prices. The range depends on whether a page returns 100 or 10 results. Firecrawl's Search endpoint bills 2 credits per 10 results. That lands between 0.12 USD (Scale) and 0.64 USD (Hobby, annual). On 2026-11-17 we called EveryInfra's `web/search` for real.

One request, 0.000694 USD, 10 results. That is 0.069 USD per 1,000. The gap is wider on full-site crawling. Firecrawl's Crawl charges 1 credit per page. So 1,000 pages is 0.60 (Scale) to 3.20 USD (Hobby). Our `crawl` returns several pages for a single 0.000694 USD request.

Two of our crawls returned 10 and 3 pages. That is 0.069 to 0.231 USD per 1,000 pages, several to dozens of times cheaper. Below we test five jobs one by one. Each comes with billing unit, failed and empty-result charging, free tiers, minimum top-up, and MCP support.

> **Key points**
>
> - Web search per 1,000 results: Apify about 0.045 to 0.45 USD, Firecrawl 0.12 to 0.64 USD, our live call 0.069 USD. All three search, and per-request is cheapest.
> - Full-site crawl per 1,000 pages: Firecrawl charges 1 credit per page, 0.60 to 3.20 USD. Our `crawl` returns several pages per 0.000694 USD request, 0.069 to 0.231 USD.
> - Single-page scrape per 1,000 pages: ours (`read`) is 0.694 USD. Firecrawl is lower on Standard and Scale (0.60 to 0.83 USD), but those need an 83 to 599 USD monthly plan whose credits expire.
> - Structured platform data per 1,000: an Apify ready-made actor is about 2.30 USD. Our `amazon/search` returned 25 rows for 0.002236 USD, which is 0.089 USD. Firecrawl has no native platform endpoint.
> - Failed and empty results: a missing key returned 401, a bad parameter returned 422, and a missing page returned 404, all charged 0. Firecrawl's page states a 403 or 404 still costs 1 credit. Our live calls cost about 0.008 USD in total.

The "per-1,000 price" divides the amount a request charged by the rows or pages it returned, then times 1,000. It answers one question: what does 1,000 results cost. EveryInfra charges per request. Firecrawl charges per credit. The scraping programs (actors) in Apify Store charge per event or per compute. The three units only compare once they share a denominator.

The denominator is always the rows actually returned, never the limit in the request. So a call that returned fewer rows only makes our side look pricier. For the other two we used public material only. That means Firecrawl's pricing page and docs, Apify's pricing page, the two actors' public metadata, and public endpoints that need no account. We did not register, buy a plan, or run any actor. So their numbers are public list prices, not measured bills.

They have nothing to do with our own data.

![Isometric illustration of three data lanes on one board. The left lane ends at a storefront of signed stalls, a marketplace of actors. The middle lane ends at a machine that folds pages into markdown. The right teal lane ends at a counter that ticks one number per press, billing per request, all drawing from one hopper of public pages.](https://everyinfra.com/articles/apify-vs-firecrawl/fig-01-three-lanes.light.svg?utm_medium=ai_doc&utm_campaign=page_md)

## Headline table: five jobs, cost per 1,000

Here are the five most common data jobs side by side. The Apify side uses public prices. Platform compute is billed per compute unit, one GB of RAM for one hour. That is 0.2 USD on Free and Starter, and 0.13 USD on Business ([Apify pricing](https://apify.com/pricing), read 2026-11-17). Store actors add their own event prices. For example, the Google Search Results Scraper charges per result page scraped.

That is 0.0045 USD per page on Free and 0.00245 on Business (Gold) ([Google Search Results Scraper metadata](https://api.apify.com/v2/acts/apify~google-search-scraper), read 2026-11-17). The Firecrawl side comes from its pricing page. One credit equals one page on a basic scrape or crawl, or one Map call. Search costs 2 credits per 10 results. Each credit runs from 0.0032 USD (Hobby annual) down to 0.000599 USD (Scale annual) ([Firecrawl pricing](https://www.firecrawl.dev/pricing), effective 2026-09-04, read 2026-11-17). The EveryInfra side is all live calls.

The amount charged comes from `charged_usd`, and the row count from the length of `results`.

| Firecrawl operation | Credits |
| --- | --- |
| Scrape | 1 per page |
| Crawl | 1 per page |
| Map | 1 per call |
| Search | 2 per 10 results |
| JSON format | +4 per page |

| Job | EveryInfra capability | Live calls: count / rows / charged (USD) | Our per-1,000 | Firecrawl public (per 1,000, annual) | Apify public (per 1,000) |
| --- | --- | --- | --- | --- | --- |
| Web search | `web/search` | 2 / 20 / 0.001388 | 0.069 | Search: 0.12 (Scale) to 0.64 (Hobby) | Google Search Scraper: 0.045 to 0.45 |
| Scrape one page | `read` | 2 / 2 / 0.001388 | 0.694 | Scrape: 0.599 (Scale) to 3.20 (Hobby) | Website Content Crawler: by compute |
| Full-site crawl | `crawl` | 2 / 13 / 0.001388 | 0.069 to 0.231 | Crawl: 0.599 (Scale) to 3.20 (Hobby) | Website Content Crawler: by compute |
| Site map | `map` | 1 / 50 URLs / 0.000694 | 0.694 (per 1,000 calls) | Map: 0.599 to 3.20 (per 1,000 calls) | Website Content Crawler: by compute |
| Structured platform data | `amazon/search` | 1 / 25 / 0.002236 | 0.089 | No native endpoint; scrape then extract | Ready-made actor about 2.30 |

![Horizontal bar chart, log scale, USD per 1,000 web search results. Apify Google Search Scraper runs 0.045 at 100 per page to 0.45 at 10 per page. Firecrawl Search is 0.12 on Scale and 0.64 on Hobby annual. EveryInfra web/search is 0.069, teal and leftmost, with its other interface at 0.139.](https://everyinfra.com/articles/apify-vs-firecrawl/fig-03-search-per1k.light.svg?utm_medium=ai_doc&utm_campaign=page_md)

Multiples always divide the other side's per-1,000 price by ours. Read the table with two things in mind. First, search, full-site crawl, and structured data each return many rows per charge. The more rows, the thinner the unit price. That is why the gap is largest there. Second, scraping one page is a single-target job.

It returns one page. There our 0.694 USD is not the bargain. Firecrawl's per-page price on Standard, Growth, and Scale is lower, at 0.599 to 0.83 USD. The catch is a prepaid 83 to 599 USD a month, and unused credits expire. Which is cheaper depends on which kind of job you have. Every section below works that out.

## How we calculate

The per-1,000 price equals the amount a request charged. Divide it by the rows returned, then times 1,000. When a job ran several times, the amounts and rows are summed first. A request that returned zero rows is left out of the denominator. It is checked on its own for whether it charged. On the Firecrawl side we turn credit rules into credits per 1,000 rows.

Scrape and crawl are 1 per page. Map is 1 per call. Search is 2 per 10 results. Then we multiply by the per-credit plan price. Plan prices use annual billing, about 16.7% below monthly. On the Apify side, platform usage is compute units times the unit price.

Store actors are read event by event. We use the last effective version in `pricingInfos` in their public metadata. A price whose start time is later than the read time is announced but not yet effective. We flag it rather than use it as current.

The live calls used a test key shared with other writers that day. So we recorded `charged_usd` directly from each response. We did not use a before-and-after balance. The key allows only 10 requests per minute. Above that it returns 429 with no charge. So the calls went out in three moments: 00:20, 00:24, and 00:38 Singapore time (SGT).

We did not fire them in parallel. Targets were stable public pages and brand or institutional keywords. No individual creator handles are named. One script recomputes every per-1,000 price, multiple, and monthly cost from the raw responses. Each number in the charts is read by the plotting script from that same result.

## Test one: web search, 10 results per call, 0.069 USD per 1,000

Web search is a job all three do. Start there. On 2026-11-17 at 00:21 Singapore time (SGT) we called `web/search` for "web scraping api pricing comparison". It returned 10 results in about 1.5 seconds. It charged 0.000694 USD. Each result carries a title, link, and snippet.

These match the fields in the catalog. To get more you page with `page`. At 00:38 Singapore time we paged to page 2. It returned 10 more rows and charged another 0.000694 USD. So a page is another request at the same fixed price. At 10 per call, 1,000 results is 100 requests, or 0.069 USD.

Here is the request:

```bash
curl -s https://api.everyinfra.com/api/v1/search \ -H "Authorization: Bearer $EVERYINFRA_API_KEY" \ -H "Content-Type: application/json" \ -d '{"tool":"web","params":{"q":"web scraping api pricing comparison","num":10,"location":"us"}}'

# returns 200, 10 rows in results, billing.charged_usd of 0.000694
```

![Screenshot of a real API response. The request is a POST with tool web, q web scraping comparison, num 10. The 200 response at 2026-11-17 00:21 Singapore time shows results length 10, with the first row and charged_usd 0.000694 outlined in teal.](https://everyinfra.com/articles/apify-vs-firecrawl/img-web-card.webp?utm_medium=ai_doc&utm_campaign=page_md)

One web search returns 10 rows for 0.000694 USD. [EveryInfra API response, web search API](/web-api/search)

Firecrawl's Search endpoint combines search and scrape in one call. By default it returns a title, link, and snippet per result at 2 credits per 10 results. Turn on scrape options in the same call and it also fetches each page as markdown. That adds the scrape credit per result ([Firecrawl Search API](https://docs.firecrawl.dev/api-reference/endpoint/search), [Firecrawl Search feature](https://docs.firecrawl.dev/features/search)). For search alone, 1,000 results is 200 credits. That is 0.12 (Scale annual) to 0.64 USD (Hobby annual).

The Free tier's 1,000 monthly credits cover 5,000 results. Firecrawl positions Search as search plus scrape in one step. It suits a research agent that reads before it answers.

Apify has no single web search endpoint. Instead the Google Search Results Scraper actor scrapes Google result pages. It now bills per event. Each result page scraped is 0.0045 USD on Free, 0.0035 on Bronze (Starter), 0.00245 on Gold (Business), and 0.0014 on Diamond. Add a 0.002 USD per-run start fee and a 0.5 USD minimum charge ([Google Search Results Scraper metadata](https://api.apify.com/v2/acts/apify~google-search-scraper), read 2026-11-17). A Google page can be set to 10 or 100 results.

At 100 per page, 1,000 results is 10 pages, 0.045 USD on Free, lower than us. At 10 per page it is 100 pages, 0.45 USD, higher.

| Apify Google Search tier | Per result page (USD) |
| --- | --- |
| Free | 0.0045 |
| Bronze (Starter) | 0.0035 |
| Gold (Business) | 0.00245 |
| Diamond | 0.0014 |

That actor has run 137 million times. It had 27,143 users in the last 30 days and a 4.65 rating. So on search the three sit in one band. Per-request puts us in the low middle.

Apify is lowest when a page packs 100 results. Firecrawl is highest on the low plans.

## Test two: scrape one page, one page per call, this one depends on commitment

Turning a URL into clean text is Firecrawl's signature job. It is the most important section against it. On 2026-11-17 at 00:21 Singapore time we called `read` on a Model Context Protocol page. It returned one page of text in about 7 seconds. The page had a title, link, and body. It charged 0.000694 USD, or 0.694 USD per 1,000 pages.

At 00:38 Singapore time we re-read the same page as a stability check. It took about 2.6 seconds, still one page, still 0.000694 USD. The text matched, so the price is steady. `read` tries several fetch sources in turn. When one cannot retrieve the page, it switches to the next. It returns cleaned body text.

Firecrawl's Scrape is 1 credit per page. It renders JavaScript by default. It returns markdown ready to feed a model. Firecrawl says that markdown uses about 67% fewer tokens than raw HTML ([Firecrawl Scrape feature](https://docs.firecrawl.dev/features/scrape), [Firecrawl: best web scraping API](https://www.firecrawl.dev/blog/best-web-scraping-api), 2026-08-11). Per 1,000 pages it runs from 3.20 (Hobby annual) through 0.83 (Standard) and 0.666 (Growth) to 0.599 USD (Scale). The Free tier covers 1,000 pages a month.

| Firecrawl plan | Monthly (USD) | Credits | Per 1,000 pages (USD) |
| --- | --- | --- | --- |
| Free | 0 | 1,000 | 0 |
| Hobby | 16 | 5,000 | 3.20 |
| Standard | 83 | 100,000 | 0.83 |
| Scale | 599 | 1,000,000 | 0.599 |

Put that next to our 0.694 USD. On Hobby we are about 4.6 times cheaper. On Standard and Scale Firecrawl's per-page price is below ours. But you must commit to 83 to 599 USD a month. And unused credits expire, except on Scale (one month) and Enterprise ([Firecrawl pricing](https://www.firecrawl.dev/pricing)). So this one depends on how steady your volume is.

Large and reliably full favors Firecrawl's higher tiers. Small, spiky, or occasional favors our 0.000694 USD per request, with no monthly fee and a balance that does not expire.

![Grouped horizontal bar chart, log scale, USD per 1,000 pages. The scrape group shows EveryInfra read 0.694 and Firecrawl Hobby 3.20, Standard 0.83, Scale 0.599. The crawl group shows EveryInfra 0.069 to 0.231 and the same Firecrawl tiers. In the crawl group the teal bar is far shorter than every Firecrawl tier. In the scrape group EveryInfra sits in the middle.](https://everyinfra.com/articles/apify-vs-firecrawl/fig-04-scrape-crawl-per1k.light.svg?utm_medium=ai_doc&utm_campaign=page_md)

On the Apify side this maps to the Website Content Crawler actor. It is built to crawl site content into markdown for models, vector databases, and RAG ([Website Content Crawler metadata](https://api.apify.com/v2/acts/apify~website-content-crawler), read 2026-11-17, last modified 2026-11-16). It bills by compute with no fixed per-page price. The cost equals the compute units a run uses times the unit price. It depends on rendering and page size. So you test-run once to learn the per-page figure.

It does more than plain text. It handles infinite scroll, login-protected pages, PDF and Office downloads, screenshots, incremental crawls, and even a per-page AI summary. It has run nearly 50 million times. It had about 12,000 users in the last 30 days and a 4.59 rating. In the public 30-day run stats, 3,612,346 succeeded, 3,109 failed, 108,485 timed out, and 21,530 were aborted. So about 3.5% did not succeed, mostly timeouts.

That tells you web scraping itself is not stable. Every provider needs retries.

## Test three: full-site crawl, several pages per call, several to dozens of times cheaper

Full-site crawl is where the gap is widest. On 2026-11-17 at 00:24 Singapore time we called `crawl` on the Model Context Protocol site. We told it in plain language to take only the docs pages about the protocol. It returned 10 pages in about 9.4 seconds. Each had a title, link, and body. The whole thing charged 0.000694 USD.

The point is here. That is the price of one request. One request returned 10 pages. So each page is 0.0000694 USD, or 0.069 USD per 1,000. At 00:38 Singapore time we crawled Firecrawl's docs site. We set `num` to 20 and asked for feature and pricing docs pages.

This time it returned 3 pages, again for 0.000694 USD, or 0.231 USD per 1,000. The two runs show one thing. How many pages a `crawl` returns depends on the site and the instruction. But whether it returns 3 or 10, it is one fixed price. So the per-page price sits between 0.069 and 0.231 USD. Both are well below Firecrawl's per-page credit.

![Screenshot of a real API response. The request is a POST with tool crawl, url modelcontextprotocol.io, num 10. The 200 response at 2026-11-17 00:24 Singapore time shows results length 10, with the first pages and charged_usd 0.000694 outlined in teal.](https://everyinfra.com/articles/apify-vs-firecrawl/img-crawl-card.webp?utm_medium=ai_doc&utm_campaign=page_md)

One crawl request returns 10 pages for 0.000694 USD. [EveryInfra API response, full-site crawl API](/products/search/crawl)

Firecrawl's Crawl charges 1 credit per page it crawls. It discovers and scrapes every subpage on a domain. It returns the site as markdown or JSON ([Firecrawl Crawl feature](https://docs.firecrawl.dev/features/crawl), [Firecrawl Crawl page](https://www.firecrawl.dev/crawl), 2026-10-29). For a 100-page site Firecrawl charges 100 credits. That is 0.32 USD on Hobby annual, 0.083 on Standard, and 0.0599 on Scale. To crawl the same 100 pages we set `num` per call and split across requests.

At 10 pages per call that is 10 requests, 0.00694 USD. That is about 12 times cheaper than Firecrawl's Standard tier and about 46 times cheaper than Hobby. The cause is the same as before. Firecrawl charges linearly per page. We charge per request. The more pages a request returns, the cheaper the page.

Here is the boundary. Our per-call page count is capped. A deep site needs several requests stitched together. Firecrawl's Crawl finishes a whole site with one command. It handles paging and deduplication for you. That is where it saves effort.

Apify again uses the Website Content Crawler, billed by compute. So the cost of deep-crawling a large site depends on the compute it uses. There is no per-page figure to compare. Its strengths are depth and detail. Incremental crawls take only changed pages. Cookies open logged-in areas.

It downloads attachments ([Website Content Crawler metadata](https://api.apify.com/v2/acts/apify~website-content-crawler)). So full-site crawl comes down to cheapest versus most hands-off. To drive cost to the floor and accept your own paging, we are cheapest. To crawl a whole site in one command with login and incremental support, Firecrawl and Apify fit better.

## Test four: site map, 50 links per call

Sometimes you only want a site's list of pages, not the body. On 2026-11-17 at 00:24 Singapore time we called `map` on the Model Context Protocol site. It returned 50 links in about 0.3 seconds. It charged 0.000694 USD. Firecrawl's Map is also URLs only and fast. But it charges 1 credit per call, the same as scraping a page ([Firecrawl Map feature](https://docs.firecrawl.dev/features/map), [Firecrawl pricing](https://www.firecrawl.dev/pricing)).

So this one is close. Both charge one price per call. Ours is 0.694 USD per 1,000 calls, Firecrawl's 0.599 to 3.20 USD. The difference is that we return 50 links per call at a fixed price. Firecrawl's link count per Map call depends on the site. A job that maps a site first, then fetches chosen pages, can run Map plus Crawl on both.

The cost is still our per-request versus Firecrawl's per-page split.

## Test five: structured platform data, the one Firecrawl cannot do

Beyond search and scrape is a different job. You pull structured fields from a specific platform. That includes Amazon product prices and ratings, or Reddit posts. On 2026-11-17 at 00:25 Singapore time we called `amazon/search` for "usb c cable". It returned 25 rows in about 54 seconds. Each had ASIN, title, brand, price, currency, rating, review count, stock, seller, and best-seller rank, 18 fields in all.

It charged 0.002236 USD, or 0.089 USD per 1,000. At the same moment `reddit/search` for "firecrawl vs apify" returned 1 row. The term is barely discussed on Reddit. It charged 0.000792 USD. Every field here is parsed, not a page of markdown. So it goes straight into price comparison, product selection, or sentiment work.

See the field notes at the [Amazon data API](/en/amazon-api/search) and [Reddit data API](/en/reddit-api/search).

![Horizontal bar chart, log scale, USD per 1,000 structured results. An Apify ready-made platform actor is about 2.30, using the Instagram Scraper example. Firecrawl has no native platform endpoint, drawn as a crossed-out placeholder. EveryInfra amazon/search is 0.089 at 25 rows per call. The teal bar is far shorter than Apify's.](https://everyinfra.com/articles/apify-vs-firecrawl/fig-06-structured-per1k.light.svg?utm_medium=ai_doc&utm_campaign=page_md)

Of the 25 rows this `amazon/search` returned, the priced items ran from a few to twenty-odd dollars. Each was tagged with brand, rating, review count, stock, an "Amazon's Choice" mark, and a best-seller rank. So product selection and price comparison can filter by field. You do not scrape a page apart. Scraping the full page instead means handling Amazon's anti-bot and layout changes. You write your own parser, with no guarantee the fields are complete.

Doing it through Firecrawl means scraping the page then extracting. JSON adds 4 credits per page. So one product page already costs more than one of our requests. For structured platform data the saving is not only money. It is the post-processing of turning markdown into fields.

This is where the three products diverge. Apify's core is this job. Store holds thousands of actors for specific sites. There are dedicated scrapers for social, e-commerce, maps, and jobs. Ready-made actor event prices commonly run 1 to 3 USD per 1,000. The pricing page's Instagram Scraper example is 2.30 USD per 1,000 on Starter ([Apify pricing](https://apify.com/pricing), [Apify Instagram Scraper](https://apify.com/apify/instagram-scraper)).

Firecrawl has no platform endpoint. To get Amazon's structured fields you scrape the product page. Then you extract markdown into JSON, at 4 credits per page ([Firecrawl Extract feature](https://docs.firecrawl.dev/features/extract), [Firecrawl pricing](https://www.firecrawl.dev/pricing)). That is both pricier and more assembly. We built common platforms into fixed capabilities. Each returns a page of structured results per request at one price.

So this class lands around 0.089 USD per 1,000.

That is one or two orders of magnitude below a ready-made actor. To be clear: our fields, Apify's fields, and Firecrawl's markdown are each built independently. No data flows between them. The fields are not the same.

## What you get back: markdown, structured fields, search snippets

Beyond price, the three return different things. Selection has to weigh that too. Firecrawl's Scrape and Crawl return a page of markdown. It strips navigation, footer, and ads, and keeps the body. It suits dropping straight into a model's context. For structure you add extraction and convert to JSON by schema.

Apify's actors follow each actor's own output. The Website Content Crawler gives markdown or plain text with metadata. The Google Search Results Scraper gives parsed search fields. Site-specific actors give their own fields. Our side splits into two. The search tools (`web`, `read`, `crawl`, `map`) return titles, links, and body or snippet.

Structured platform capabilities return parsed fields. Examples are the [Amazon data API](/en/amazon-api) and the [Reddit data API](/en/reddit-api). The table below lays out which is which.

Match format to need. Whole-page body for a model reads best as markdown, from Firecrawl Scrape and Crawl or our `read` body. Fields for a database come parsed, from our platform capabilities or an Apify site actor. Search then fetch on demand uses snippets plus a fetch, with our `web` plus `read` or Firecrawl Search. A specific obscure site needs that site's actor from Apify Store. JSON from any page by schema needs an extraction add-on, which on Firecrawl Extract costs 4 extra credits.

So "which is cheaper" has to be asked with "which format". For a product's price and rating, Firecrawl means scraping a page then spending 4 credits to extract JSON. Our structured capability returns the fields in one request. For a document's full text into RAG, Firecrawl's markdown is directly usable. Our `read` cleaned body is enough. But Firecrawl fetches a reflowed, front-end-rendered page more completely.

Put the format need into the cost. Then you will not pick the wrong tool on unit price alone.

## Success rates: Apify's public run stats

Apify publishes each actor's last-30-day run results in its metadata. You can read them directly. The Website Content Crawler ran 3,745,470 times in the last 30 days. Of those, 3,612,346 succeeded, 3,109 failed, 108,485 timed out, and 21,530 were aborted. So about 3.5% did not succeed, mostly timeouts ([Website Content Crawler metadata](https://api.apify.com/v2/acts/apify~website-content-crawler)). The Google Search Results Scraper ran 9,742,552 times.

Of those, 9,551,726 succeeded, 440 failed, 171,838 timed out, and 18,548 were aborted. That is about 2.0% not succeeding ([Google Search Results Scraper metadata](https://api.apify.com/v2/acts/apify~google-search-scraper)). Both ratios say web scraping itself is unstable. Timeouts are normal. Whichever provider you use, you need retries and alerting. Do not trust a "run succeeded" alone.

That flags something. A per-1,000 price is computed on successfully returned rows. It leaves out the cost of retries. On a pay-per-event platform a failed run usually produces no result event. But a miss still means a rerun and another wait. We charge per request.

A failure is not charged. A retry is just another request. Factor retries in. An unstable target raises cost on both models. Nobody avoids that piece. The only difference is whether the failed attempt charged.

Third-party benchmarks now weigh success rate with cost ([Firecrawl: best web search APIs](https://www.firecrawl.dev/blog/best-web-search-apis), 2026-10-21; [Firecrawl reviews on G2](https://www.g2.com/products/firecrawl/reviews), read 2026-11-15).

## Break-even: how many rows per request before per-request wins

The line between the two billing styles comes from one number. Our request is 0.000694 USD. The other side charges per page or per row. Divide our per-call price by their per-row price. That gives how many rows a request must return before we are cheaper.

| Other side per 1,000 | Example | Break-even rows |
| --- | --- | --- |
| 3.20 USD | Firecrawl Scrape, Hobby | 0.22 |
| 0.83 USD | Firecrawl Standard | 0.84 |
| 0.599 USD | Firecrawl Scale | 1.16 |
| 0.45 USD | Apify 10 per page Free | 1.54 |
| 0.045 USD | Apify 100 per page Free | 15.4 |

Read the break-even column plainly. Against Firecrawl's Hobby tier, even one row makes us cheaper. Against Standard, one row breaks even and two wins. Against Scale and Apify at 10 per page, two or more rows wins. Against Apify at 100 per page, you need 16 or more rows.

The table ties the results together. Scraping one page returns one page. So it loses to Firecrawl's Scale tier and Apify's 100-per-page math. Search returns 10. Crawl returns several pages. Structured returns dozens.

All are far past break-even, so the gap is wide. Before switching, count how many rows a request returns on average. That number decides the answer more than any list price.

![Line chart, x-axis rows per request from 1 to 100 on a log scale, y-axis USD per 1,000. The teal EveryInfra line falls from 0.694 at one row to 0.0069 at 100 rows. Grey lines mark fixed tiers: Firecrawl Hobby 3.20, Standard 0.83, Scale 0.599, and Apify 100-per-page 0.045. Break-even rows are marked where the teal line crosses each grey line.](https://everyinfra.com/articles/apify-vs-firecrawl/fig-05-breakeven.light.svg?utm_medium=ai_doc&utm_campaign=page_md)

## A worked month: three common jobs

Here is a month (30 days) for three common uses. Firecrawl takes the cheapest plan that fits the volume, at the annual price. Apify takes the ready-made actor event price plus a plan fee. EveryInfra equals requests times 0.000694 USD. Structured capabilities use their own unit price.

| Monthly job | Rows | Firecrawl | Apify | EveryInfra |
| --- | --- | --- | --- | --- |
| RAG: 200-page docs daily | 6,000 | 16 USD | by compute | 0.42 USD |
| 20 keywords, 100 each daily | 60,000 | 83 USD | 2.70 USD | 4.17 USD |
| 1,000 product pages daily | 30,000 | scrape+extract | 69 USD | 2.68 USD |

![Dumbbell chart, USD per month on a log scale, one row per job. Docs-site RAG: EveryInfra 0.42, Firecrawl 16, Apify by compute. Keyword search: EveryInfra 4.17, Apify 2.70, Firecrawl 83. Product structured: EveryInfra 2.68, Apify 69, Firecrawl not native. EveryInfra's points are teal.](https://everyinfra.com/articles/apify-vs-firecrawl/fig-08-monthly-cost.light.svg?utm_medium=ai_doc&utm_campaign=page_md)

The first row needs a note. For docs-site RAG, Firecrawl's Crawl finishes in one command at 1 credit per page. 6,000 pages just overflows Hobby's 5,000 credits, so you top up or move a tier. We estimate 600 requests at 10 pages each, about 0.42 USD. The second row is search. Apify's 100-per-page is cheapest at volume.

We are next. Firecrawl is priciest because the plan starts at 83 USD a month. The third row is structured product data. Firecrawl has no native endpoint and means scrape-then-extract at 4 extra credits per page. Apify's ready-made actor is clear but costly. We are cheapest.

One more reminder. Firecrawl and Apify plan fees are prepaid usage. They do not roll over and expire at the period's end. Our balance does not expire ([Firecrawl pricing](https://www.firecrawl.dev/pricing), [Apify Console Billing](https://docs.apify.com/platform/console/billing)).

## Billing unit, failed charges, free tiers: three models

Side by side, the three are three models. EveryInfra charges per request at one fixed price. The balance never expires. Firecrawl charges per credit. It sells monthly credits with a plan and 5 USD top-ups. Credits expire at month end, except on Scale and Enterprise.

Apify charges per event or per compute. Prepaid plan usage does not roll over and expires ([Apify pricing](https://apify.com/pricing), [Firecrawl pricing](https://www.firecrawl.dev/pricing)).

Whether failed and empty results charge is the most overlooked line. It moves a real bill the most. We tested three "no data" cases. On 2026-11-17 at 00:20 Singapore time a request with no key returned 401 with "Nothing was charged". At the same moment a mistyped parameter returned 422. It listed the allowed parameters and a working example, with no charge.

At 00:26 Singapore time a `read` on a missing page returned 404. It had `charged` false and the reason `request_failed_refunded`, charged 0. None charged. Firecrawl's pricing page is plain. A scrape with no result is not charged. But a page that responds with 403 or 404 is still returned and costs 1 credit ([Firecrawl pricing](https://www.firecrawl.dev/pricing)).

On an Apify pay-per-event actor, whether an error record charges depends on the author's event definition. You read it from the run's event counts afterwards ([Apify pay-per-event](https://docs.apify.com/actors/publishing/monetize/pay-per-event)).

![Screenshot of a real API response. The request is a POST with tool read and a non-existent url. The HTTP 404 response at 2026-11-17 00:26 Singapore time shows error target_not_found, charged false, and reason request_failed_refunded, outlined in teal.](https://everyinfra.com/articles/apify-vs-firecrawl/img-read404-card.webp?utm_medium=ai_doc&utm_campaign=page_md)

Reading a missing page returns 404 with zero charge. [EveryInfra API response, scrape API](/products/search/read)

Free tiers and minimum top-ups differ too. Firecrawl gives 1,000 credits a month with no card. That is about 1,000 pages or 5,000 searches. It also has a no-signup, per-IP Keyless entry to try directly ([Firecrawl: best web scraping API](https://www.firecrawl.dev/blog/best-web-scraping-api)). Apify's Free plan gives 5 USD of platform usage a month with no card. We give a one-time trial credit for starring two open-source repos in the console.

Paid use starts from a top-up, and the balance does not expire. Together: occasional trials fit all three free tiers. For a long run you match the billing model to your volume and spikiness.

![Comparison figure, three columns for three providers, each with three failure cases. EveryInfra shows no key 401, bad parameter 422, and missing page 404, all 0 USD with a teal check. Firecrawl shows no result not charged, but 403 or 404 costs 1 credit, with a yellow warning. Apify shows a question mark, noting it depends on the event definition.](https://everyinfra.com/articles/apify-vs-firecrawl/fig-07-failed-charge.light.svg?utm_medium=ai_doc&utm_campaign=page_md)

![Screenshot of the Firecrawl pricing page, section Pricing at a glance. It states 1 credit per page on scrape or crawl, 1 per Map call, and 2 per 10 search results. It states a free tier of 1,000 credits a month, and that 403 or 404 costs 1 credit. The footer reads Effective September 4, 2026.](https://everyinfra.com/articles/apify-vs-firecrawl/img-firecrawl-pricing.webp?utm_medium=ai_doc&utm_campaign=page_md)

Firecrawl pricing page credit rules, read 2026-11-17. [Firecrawl pricing page](https://www.firecrawl.dev/pricing)

Here are the five dimensions in plain prose. On billing unit, EveryInfra charges per request at one price. Firecrawl charges per credit, by page, call, or per 10 results. Apify charges per event or per compute unit. On failed or empty results, our 401, 422, and 404 all charged 0 in tests. Firecrawl does not charge a no-result scrape, but 403 and 404 cost 1 credit.

Apify depends on the actor's event definition. On monthly fee, we have none and the balance does not expire. Firecrawl runs 16 to 599 USD a month, and most credits expire. Apify runs 19 to 999 USD a month, with no roll over. On free tier, we give a one-time trial for starring repos. Firecrawl gives 1,000 credits a month plus a no-signup Keyless.

Apify gives 5 USD of platform usage a month. On MCP, we are a remote MCP with tools named everyinfra_*. Firecrawl's server is firecrawl. Apify's is mcp.apify.com.

## Calling from AI tools: all three speak MCP

All three offer MCP. That is the Model Context Protocol, the protocol an AI client uses to call external tools ([2025-06-18 spec](https://modelcontextprotocol.io/specification/2025-06-18)). EveryInfra is a remote MCP. Its tool names all start with `everyinfra_`. You add one block in a client's config:

```text
{ "mcpServers": { "everyinfra": { "url": "https://api.everyinfra.com/mcp", "headers": { "Authorization": "Bearer ${env:EVERYINFRA_API_KEY}" } } } }

```

Firecrawl's MCP server is named `firecrawl`. It supports no-signup Keyless, account sign-in, or an API key. It exposes scrape, crawl, search, map, and extract as tools ([Firecrawl MCP docs](https://docs.firecrawl.dev/mcp-server), [firecrawl-mcp-server source](https://github.com/firecrawl/firecrawl-mcp-server)). Apify's MCP is at mcp.apify.com. It recommends OAuth sign-in. A model can search Store for an actor, then run it ([Apify MCP docs](https://docs.apify.com/platform/integrations/mcp), [apify-mcp-server source](https://github.com/apify/apify-mcp-server)).

Use in an AI client is similar across the three. The difference is still billing. The more a model fetches, Firecrawl adds credits, Apify adds events, we add requests. Say in chat: search the top 10 for this term, then scrape the first result. The model calls `web/search`, then one fetch. That is two requests at 0.000694 USD each.

Setup details are in our [remote MCP acceptance guide](/en/build/remote-mcp-setup).

## Migrating over: what to change

Once the price is clear, the migration cost is in the engineering details. From Firecrawl, the first change is the unit. Firecrawl's Scrape and Crawl work in pages and return markdown. Our search capabilities work in requests and return structured fields or body text. A script that parsed markdown switches to reading fields. The second is crawl paging.

Firecrawl finishes a whole site in one Crawl command. We set `num` high and split across requests. The third is format. Firecrawl gives rendered markdown by default, ideal to feed a model. Our `read` gives cleaned body text. It may miss parts of a heavily reflowed single-page app.

Test one page when unsure.

From Apify, the main change is the call shape. You replace "start a run, then fetch the dataset" with "send one request, get the result". An Apify run processes a batch of inputs into a dataset. Unnamed datasets have a retention window ([Apify Dataset docs](https://docs.apify.com/platform/storage/dataset)). We return a single target synchronously per request. Error handling differs too.

Apify's error is a record with an error code in the dataset. You filter it out. Ours is an HTTP status plus an error code. 401, 422, and 404 are not charged. An empty result states a refund reason in the response. On rate, each account defaults to some requests per minute shared across keys.

It speeds up with top-ups. Size the rate before a large run. Compare the two sides at [Apify platform limits](https://docs.apify.com/platform/limits) and [Apify usage and resources](https://docs.apify.com/platform/actors/running/usage-and-resources).

![Flow diagram with one teal dot around six stations. Station 1 sends a request; station 2 checks the key and params, returning 401 or 422 with no charge. Station 3 fetches, synchronously or async; station 4 charges one price for data and refunds a missing page. Station 5 returns a list, body, or several pages, and station 6 records one row per request.](https://everyinfra.com/articles/apify-vs-firecrawl/fig-02-request-flow.light.svg?utm_medium=ai_doc&utm_campaign=page_md)

## When Firecrawl fits better

It is easy to write a comparison that only flatters your own side. Here are cases where Firecrawl fits better, or where we cannot yet. First, turning many modern sites into clean markdown for a model. These are React, Vue, and other front-end-rendered single-page apps. Firecrawl renders JavaScript by default and returns LLM-friendly markdown. It says that uses about 67% fewer tokens than raw HTML.

That is a real win for a RAG pipeline. Second, interacting while scraping. That means clicking "load more", filling a form, or paging before fetching. Firecrawl's Interact feature does this in natural language or code ([Firecrawl Interact feature](https://docs.firecrawl.dev/features/interact)). Third, extracting pages to structured JSON by schema, or monitoring page changes. Firecrawl's extraction and Monitor do these, each at extra credits.

Fourth, large and steady volume with monthly prepayment. Standard and Scale per-page prices beat ours. Fifth, enterprise zero data retention, SSO, and IP and key allowlists. Firecrawl's enterprise plan provides these ([Firecrawl enterprise](https://www.firecrawl.dev/enterprise)). Firecrawl raised a 75 million USD Series B and launched Alexandria in September 2026 ([Wowtale](https://en.wowtale.net/2026/09/23/235198/), 2026-09-23). So it iterates fast.

## When Apify fits better

Apify's cases are just as clear. First, when your target is a specific and obscure site. Store's thousands of actors cover long-tail sites far beyond our fixed capabilities. Third-party developers keep adding more ([Apify actors in Store](https://docs.apify.com/actors/running/actors-in-store)). Second, deep-crawling a large site with login, incremental updates, attachment downloads, screenshots, and per-page AI summaries. The Website Content Crawler does this more fully than we do.

Third, built-in schedules, webhooks, and ready cloud-storage and automation integrations. Apify's console is more complete. Concurrency is wide, up to 256 runs at once on Business. Fourth, being an autonomous agent with a crypto wallet that pays per use with no account. Apify supports this through x402 and MPP ([Apify pricing](https://apify.com/pricing)). We fit the opposite.

One class of data needs many rows with only the latest page each time. Cost has to hit the floor. You want fixed structured fields. You care that "no data means no charge" is hard-coded and visible. You want search and crawl wired straight into an AI tool.

## Recent changes (September to November 2026)

| Date | Event | What it means for you |
| --- | --- | --- |
| 2026-11-16 | Apify's Website Content Crawler and Google Search Results Scraper metadata both updated this day | actor inputs and prices change any time; read the day's public metadata when comparing |
| 2026-11-11 | An agent search API benchmark put Firecrawl next to Exa, Tavily, and Perplexity on accuracy, cost, and latency ([AIMultiple](https://aimultiple.com/agentic-search)) | search API selection increasingly weighs total cost, not unit price alone |
| 2026-11-05 | A third-party comparison framed Firecrawl as stronger on site-wide ingestion and configurable extraction ([TinyFish](https://www.tinyfish.ai/blog/tinyfish-vs-firecrawl)) | Firecrawl's strength is site-wide and extraction, not the lowest unit price |
| 2026-11-02 | Firecrawl refreshed its best web extraction tools 2026 list, comparing itself with Apify ([Firecrawl blog](https://www.firecrawl.dev/blog/best-web-extraction-tools)) | both treat the other as a competitor, features converging |
| 2026-10-29 | Firecrawl's Crawl page updated, stressing one command for a whole site ([Firecrawl Crawl](https://www.firecrawl.dev/crawl)) | full-site crawl is Firecrawl's headline scenario |
| 2026-10-24 | A third party computed per-1,000-page cost under Firecrawl's 5-credit JSON rule ([WebCrawlerAPI](https://webcrawlerapi.com/blog/firecrawl-pricing)) | price the extra credits before turning on JSON or extraction |
| 2026-09-23 | Firecrawl raised a 75 million USD Series B and launched Alexandria ([Wowtale](https://en.wowtale.net/2026/09/23/235198/)) | funding and team expanding, faster iteration |
| 2026-09-04 | Firecrawl's current pricing took effect: 1 credit per page, Search 2 credits per 10 results, 403/404 costs 1 credit ([Firecrawl pricing](https://www.firecrawl.dev/pricing)) | current prices are this version; older articles may be stale |

## Compliance boundary: public data, site terms, personal data

All three fetch public web pages. But scraping is bound by two layers. One is each site's own terms. Many platforms prohibit unauthorized automated scraping. Read the target site's terms and robots rules first. The other is personal-data law.

What you fetch may contain names, comments, or contact details. In the EU you must handle it under the General Data Protection Regulation ([Regulation (EU) 2016/679](https://eur-lex.europa.eu/eli/reg/2016/679/oj)). Store and use it under your local law. Firecrawl's enterprise plan offers zero data retention. It can strip personal information from output ([Firecrawl enterprise](https://www.firecrawl.dev/enterprise)). Apify's actor pages note results may contain personal data protected by law.

We read only public pages. We cannot reach private content. Personal data in results must be handled under local law. All three tools only fetch public pages back. The compliance duty sits with the user. Confirm the target site allows it before fetching.

Follow local law on use and retention after. That is independent of which provider you use. This article covers data method and cost only. It is not legal advice.

## How we verified these numbers

Every number traces to a raw response. First, the other side's prices come from public endpoints and pricing pages, not marketing copy. Apify's per-event prices come from the last effective version of `pricingInfos` in both actors' metadata. Firecrawl's credit rules and failed-charge note come from the Pricing at a glance block. Each has the read date. Second, our charge per request comes from `charged_usd` in the response.

The row count comes from the length of the `results` array, not the limit. Third, one script recomputes every per-1,000 price, multiple, and monthly cost from the raw responses. Each number in the charts is read from that result and checked against the text. Fourth, failed and empty charging is verified on three real cases: no key, a mistyped parameter, and a missing page. All had `charged` false.

We also state what we did not do. We did not register with Firecrawl or Apify. We did not buy a plan or run any actor. So their side has public list prices only. There are no measured bills, success rates, or data quality. Our own live calls cover search, scrape, crawl, site map, and two structured capabilities.

They do not cover every capability in the catalog. How many pages a `crawl` returns depends on the site and instruction. We tested only 3 and 10 pages. A deeper site needs several tests to settle a per-page price. These boundaries do not change the conclusion. Per-request billing is cheaper on data that returns in batches.

But recompute with your own returned row counts rather than copying ours.

## Rate, concurrency, and batch limits

Before a large run, rate and concurrency decide how much you move an hour. They matter as much as unit price. Firecrawl's rate is per plan and per endpoint. The table below sets the three side by side.

The three rate models differ by what they limit. Firecrawl limits per endpoint. Free allows 10 scrapes and 2 crawls a minute with 2 browsers. High tiers reach 500 to 10,000 a minute with up to 100 browsers. Apify limits concurrent runs, and one run is a batch. Free allows 5 runs, Starter 32, and Business 256.

EveryInfra limits per request, shared across keys. Free gives some per minute with auto speed-up, and top-ups raise the limit.

These three rates do not compare directly. The unit differs. Firecrawl limits per endpoint and fetches one page at a time ([Firecrawl rate limits](https://docs.firecrawl.dev/rate-limits)). Apify limits per run, and a run fetches a batch ([Apify platform limits](https://docs.apify.com/platform/limits)). We limit per request, and a request returns a page or a batch. Translate them into rows per hour.

First know whether your job scrapes page by page or in batches. Page by page, Firecrawl's high tiers at hundreds to thousands a minute are fastest. In batches, one Apify run handles hundreds of targets, so concurrency is not the bottleneck. Returning a page or dozens of rows per request, we suit spreading targets across many requests. Before a large launch, put rate, concurrency, and per-1,000 price in one table. That is more accurate than price alone.

## Getting started: from one trial to steady use

First, get an API key. Register in the console. Star two open-source repos to claim a one-time trial credit. It covers one to two thousand requests at 0.000694 USD each. Paid use starts from a top-up, and the balance does not expire ([quickstart](/en/docs#quickstart), [pricing](/en/pricing)). Second, pick one job and run it once.

Use `web/search` for web search. Use `read` to scrape a page. Use `crawl` with `num` set to your page count to crawl a site. Third, read `charged_usd` and the row count in the response. Compute your scenario's per-1,000 price as charge divided by rows times 1,000. Compare it with your current Firecrawl or Apify bill.

Common capabilities are at [web search](/en/web-api/search). Structured platform data is at the [Amazon data API](/en/amazon-api/search) and [Reddit data API](/en/reddit-api/search). Or browse by platform in the [data market](/en/market). If you already use Firecrawl or Apify, see our [Firecrawl alternatives](/en/blog/firecrawl-alternatives) and [Apify alternatives overview](/en/blog/apify-alternatives).

## Frequently asked questions

### What is the difference between Apify and Firecrawl, and which should I pick?

From the Google suggestion "apify or firecrawl". The two are positioned differently. Firecrawl turns any page into model-ready markdown. It renders JavaScript. It suits scrape, crawl, and search-plus-extract in one step, and bills per credit. Apify is a marketplace of scraping programs.

It has thousands of actors for specific sites, the widest coverage, billing per event or compute. Research, RAG, and clean-markdown jobs lean Firecrawl. Scraping specific obscure sites with depth leans Apify. For fixed structured fields plus web search at the lowest cost, add EveryInfra as a third option.

### How much is one Firecrawl scrape, and is the free tier enough?

From the suggestion "firecrawl pricing". Firecrawl's basic scrape and crawl are 1 credit per page. Map is 1 per call. Search is 2 per 10 results. Advanced formats like JSON add 4 credits per page ([Firecrawl pricing](https://www.firecrawl.dev/pricing)). Each credit runs from 0.0032 USD on Hobby annual to 0.000599 on Scale.

The free tier is 1,000 credits a month with no card. That is about 1,000 pages or 5,000 searches. It is enough for occasional use. A long run needs a paid plan. Note that most plans' credits expire at month end.

### How does the Apify Google Search Results Scraper bill for search?

From the suggestion "apify google search scraper". It bills per event. Each result page scraped is 0.0045 USD on Free, 0.0035 on Starter (Bronze), 0.00245 on Business (Gold), and 0.0014 on Diamond. Add a 0.002 USD per-run start fee and a 0.5 USD minimum ([Google Search Results Scraper metadata](https://api.apify.com/v2/acts/apify~google-search-scraper)). A page can hold 10 or 100 results. So 1,000 results is about 0.045 USD at 100 per page (Free), and about 0.45 at 10 per page.

Our `web/search` is 0.000694 USD for up to 10 results, 0.069 USD per 1,000.

### Why is full-site crawl dozens of times apart but single-page scrape close?

From this article's tests. In full-site crawl, Firecrawl charges 1 credit per page. Our `crawl` returns several pages per request at one charge. So the more pages per call, the larger the gap, dozens of times at 10 pages here. Scraping one page returns one page. There our per-request 0.694 USD is close to Firecrawl's high-tier per-page price of 0.599 to 0.83 USD, where Firecrawl is actually lower.

So check how many rows a request returns first. Many, and per-request wins. One target at a time, and you count the plan fee too.

### Will a failed scrape or a missing page charge me?

From the suggestion "does firecrawl charge for failed requests". In our tests, no key returned 401, a bad parameter returned 422, and a missing page returned 404. All charged 0, with the refund reason in the response. Firecrawl's rule is that a scrape with no result is not charged. But a page returning an error status such as 403 or 404 is still returned and costs 1 credit ([Firecrawl pricing](https://www.firecrawl.dev/pricing)). On an Apify pay-per-event actor, whether an error record charges depends on the author's event definition.

You read it from the run's event counts. Before a large run, this rule moves the final bill more than list price.

### Can Firecrawl and Apify plug into Claude, Cursor, and other AI tools?

From the suggestion "firecrawl mcp". Yes, all three offer MCP. Firecrawl's server is `firecrawl`. It supports no-signup Keyless, account, or API key ([Firecrawl MCP docs](https://docs.firecrawl.dev/mcp-server)). Apify's is at mcp.apify.com. After OAuth, a model can search Store for an actor and run it ([Apify MCP docs](https://docs.apify.com/platform/integrations/mcp)).

We are a remote MCP with tools named `everyinfra_*`. You add one URL and key. The difference in an AI tool is still billing. The more a model calls, each adds cost by its own unit.

## References

1. Firecrawl pricing (credit rules, plans, failed charging), Firecrawl, effective 2026-09-04, read 2026-11-17. https://www.firecrawl.dev/pricing
1. Firecrawl Search API (search plus scrape, fields), Firecrawl docs, read 2026-11-17. https://docs.firecrawl.dev/api-reference/endpoint/search
1. Firecrawl Search feature, Firecrawl docs. https://docs.firecrawl.dev/features/search
1. Firecrawl Scrape feature (rendering, markdown output), Firecrawl docs. https://docs.firecrawl.dev/features/scrape
1. Firecrawl Crawl feature (full-site crawl, per-page billing), Firecrawl docs. https://docs.firecrawl.dev/features/crawl
1. Firecrawl Map feature (URLs only), Firecrawl docs. https://docs.firecrawl.dev/features/map
1. Firecrawl Extract feature (schema extraction, JSON format), Firecrawl docs. https://docs.firecrawl.dev/features/extract
1. Firecrawl Interact feature (click, fill, page), Firecrawl docs. https://docs.firecrawl.dev/features/interact
1. Firecrawl rate limits, Firecrawl docs. https://docs.firecrawl.dev/rate-limits
1. Firecrawl Crawl API, Firecrawl docs. https://docs.firecrawl.dev/api-reference/endpoint/crawl-post
1. Firecrawl MCP get started (Keyless, account, API key), Firecrawl docs, read 2026-11-17. https://docs.firecrawl.dev/mcp-server
1. firecrawl-mcp-server source, GitHub. https://github.com/firecrawl/firecrawl-mcp-server
1. Pay-per-event pricing model (whether error records charge), Apify docs. https://docs.apify.com/actors/publishing/monetize/pay-per-event
1. Actors in Apify Store (per-event and per-compute models), Apify docs. https://docs.apify.com/actors/running/actors-in-store
1. Apify MCP server, Apify docs. https://docs.apify.com/platform/integrations/mcp
1. apify-mcp-server source, GitHub. https://github.com/apify/apify-mcp-server
1. Apify Console Billing (prepaid usage, overage invoicing), Apify docs. https://docs.apify.com/platform/console/billing
1. Apify Usage and resources (compute units, storage, proxies), Apify docs. https://docs.apify.com/platform/actors/running/usage-and-resources
1. Apify Dataset (retention, export), Apify docs. https://docs.apify.com/platform/storage/dataset
1. Apify Platform limits (concurrency, rate), Apify docs. https://docs.apify.com/platform/limits
1. Apify n8n integration, Apify docs. https://docs.apify.com/platform/integrations/n8n
1. Get Actor endpoint (read an actor's public info), Apify docs. https://docs.apify.com/api/v2/actor-get
1. Actor runs and builds (run status definitions), Apify docs. https://docs.apify.com/platform/actors/running/runs-and-builds
1. Model Context Protocol specification 2025-06-18. https://modelcontextprotocol.io/specification/2025-06-18
1. Regulation (EU) 2016/679 (GDPR), EUR-Lex. https://eur-lex.europa.eu/eli/reg/2016/679/oj
1. EveryInfra capability catalog (price_usd, max_limit, fields), read 2026-11-17. https://api.everyinfra.com/api/v1/social/catalog
1. EveryInfra search tools catalog (tool, price, params), read 2026-11-17. https://api.everyinfra.com/api/v1/search/tools
1. EveryInfra llms.txt (failed and empty results not charged). https://api.everyinfra.com/llms.txt
1. Firecrawl Crawl page (one command for a whole site), Firecrawl, 2026-10-29. https://www.firecrawl.dev/crawl
1. 8 Best Web Scraping APIs in 2026 (Keyless, 67% tokens), Firecrawl blog, 2026-08-11. https://www.firecrawl.dev/blog/best-web-scraping-api
1. Best Web Extraction Tools for AI in 2026 (compared with Apify), Firecrawl blog, 2026-11-02. https://www.firecrawl.dev/blog/best-web-extraction-tools
1. Firecrawl enterprise (zero data retention, SSO, IP allowlist), Firecrawl. https://www.firecrawl.dev/enterprise
1. Apify pricing (compute, plans, proxies, storage, Store discounts, x402), Apify, read 2026-11-17. https://apify.com/pricing
1. Website Content Crawler public metadata (billing model, run stats, modified date), Apify API, read 2026-11-17. https://api.apify.com/v2/acts/apify~website-content-crawler
1. Google Search Results Scraper public metadata (per-event prices, run stats), Apify API, read 2026-11-17. https://api.apify.com/v2/acts/apify~google-search-scraper
1. Apify Instagram Scraper (example price), Apify. https://apify.com/apify/instagram-scraper
1. Agentic Search API Benchmark (Firecrawl vs Exa, Tavily, Perplexity), AIMultiple, 2026-11-11. https://aimultiple.com/agentic-search
1. TinyFish vs Firecrawl (Firecrawl strong on site-wide and extraction), TinyFish, 2026-11-05. https://www.tinyfish.ai/blog/tinyfish-vs-firecrawl
1. Firecrawl Pricing vs WebCrawlerAPI: Cost per 1000 Pages, WebCrawlerAPI, 2026-10-24. https://webcrawlerapi.com/blog/firecrawl-pricing
1. Firecrawl Review: Costs, Limits, and What I'd Use Instead, thunderbit, 2026-11-11. https://thunderbit.com/blog/firecrawl-review-and-alternatives
1. Firecrawl Raises 75M Series B, Launches Alexandria, Wowtale, 2026-09-23. https://en.wowtale.net/2026/09/23/235198/
1. Best Web Search APIs for AI Applications in 2026 (success rate and cost together), Firecrawl blog, 2026-10-21. https://www.firecrawl.dev/blog/best-web-search-apis
1. Firecrawl Reviews 2026 (real user reviews), G2, read 2026-11-15. https://www.g2.com/products/firecrawl/reviews
