Spicrawl vs Crawl4AI: a hosted API or an open-source crawler?
Choose Spicrawl for a hosted API with free cache hits and nothing to run. Choose Crawl4AI for open source, self-hosting, site crawling and stealth.
Crawl4AI is the most popular open-source crawler for LLMs: a free Python library, under Apache-2.0, that you run on your own machines. It now also has a paid cloud that runs the same engine behind an API key. Spicrawl is a hosted scraping API: send a URL, get Markdown or JSON back, with nothing to install. Here is a fair side-by-side, including where Crawl4AI is the better choice.
Open source
Crawl a whole site
Cached results
Facts about Crawl4AI were checked against its own website and docs on 1 October 2026. See sources.
Which one should you choose?
Choose Spicrawl if you
- want a hosted API with nothing to install or run
- want cache hits to cost nothing
- work in JavaScript or TypeScript
- want a free beta, no credit card
Choose Crawl4AI if you
- want open source you can run and change
- need to crawl whole sites or follow links
- need stealth mode or managed proxies today
- want LLM extraction with your choice of model
Spicrawl vs Crawl4AI at a glance
marks the stronger option for that row. No mark means it is even, or it depends on your needs.
| Feature | Spicrawl | Crawl4AI |
|---|---|---|
| What it is | SpicrawlA hosted scraping APIA web scraping API. You send a URL and get the page back as Markdown, HTML, text, PDF or JSON. | Crawl4AIPython library + cloud APIAn open-source Python library that drives a browser on your machine, plus Crawl4AI Cloud, a hosted API that runs the same engine. |
| Price today | SpicrawlFree during betaFree while in beta, with no credit card. Prices after the beta have not been announced. | Crawl4AILibrary free; cloud creditsThe library is free. The cloud gives $10 of credit to start with no card, plus $5 a month, then credit packs of $10, $25 or $100. This is launch pricing and may change. |
| What you run yourself | SpicrawlNothingNothing. Spicrawl is a hosted service: browsers, scaling and the cache run on its side. | Crawl4AIBrowsers, unless cloudWith the library, you install and run the browsers, proxies and servers, or its Docker image. The cloud removes that: no browser to install and no proxies to buy. |
| How requests are billed | Spicrawl1, 3 or 8 creditsPer successful request, by the engine that ran: 1 credit for a plain request, 3 for browser rendering, 8 for full Chromium. | Crawl4AICloud: 0.2 to 4 creditsThe library has no per-request charge. On the cloud a credit is $0.001: a page is 0.2 credits, a real browser 2 and a hard site 4. LLM extraction adds token costs. |
| Blocked and failed requests | SpicrawlFree, incl. 403sErrors, bot challenges and timeouts cost 0. A page that answers with a status such as 403 also costs 0. | Crawl4AICloud: not chargedOn the cloud, a blocked page, a site that does not answer, a 5xx and a request it refuses itself are not charged. |
| Cached results | SpicrawlStronger option: FreeFree. The cache is on by default and keeps results for up to 48 hours. | Crawl4AICloud: half priceOn the cloud, a result from its archive costs half the normal price. The library’s own cache is free, but runs on your machine. |
| Output formats | SpicrawlMarkdown, HTML, JSON, PDFMarkdown, HTML, text, PDF, JSON and screenshots. | Crawl4AIMarkdown, HTML, PDF +moreRaw and filtered (“fit”) Markdown, HTML, JSON, screenshots and PDFs, with links turned into numbered citations if you want them. |
| Structured data | SpicrawlSelectors + schema freePage metadata, CSS selector maps and JSON Schema at no extra cost. Extraction with a model prompt is coming soon. | Crawl4AIStronger option: CSS/XPath + LLM extractionCSS and XPath schemas, plus LLM extraction with any model LiteLLM supports, using your own API key. The cloud’s /extract works without your own key. |
| Browser actions | SpicrawlUp to 50 stepsClick, type, select, scroll, wait and run JavaScript, up to 50 steps per request, on the free beta. | Crawl4AIRun JS, wait, sessionsRun JavaScript, wait for a selector or condition, and reuse a session across calls to click “Load more” or fill forms. Virtual scrolling for feeds. |
| Crawl a whole site | SpicrawlNoNo. Spicrawl does not follow links for you: you send the URLs, up to 10,000 per batch call. | Crawl4AIStronger option: Yes, BFS, DFS, best-firstYes, in the library. Breadth-first, depth-first and best-first deep crawls with depth, page and URL filters, plus URL discovery from sitemaps. |
| Many URLs | SpicrawlUp to 10,000 URLsBatch API: up to 10,000 URLs per call, run in parallel, with results as JSON Lines kept for 72 hours. | Crawl4AIarun_many; cloud 10,000arun_many runs a list of URLs in the library. The cloud streams batches of up to 50, and background jobs take up to 10,000 URLs. |
| Proxies | SpicrawlBYO, no surchargeBring your own proxy with no surcharge. Managed proxies are coming soon. | Crawl4AIStronger option: BYO + rotation; cloud managedThe library uses your own proxies, with round-robin rotation. The cloud offers ISP and residential proxies with a choice of country. |
| Stealth and bot walls | SpicrawlNo stealth in betaNo stealth mode during the beta. mode=auto tries a plain request first and uses a browser only when needed. | Crawl4AIStronger option: Stealth mode; cloud handlesThe library has a stealth mode with an undetected-browser adapter and can connect to a remote browser over CDP. The cloud handles bot walls for you. |
| Staying logged in | SpicrawlSessionsSessions keep cookies, browser storage and the same exit IP between requests. | Crawl4AIBrowser profilesPersistent browser profiles keep cookies and storage, so you log in once and crawl as that user. You manage the profile folders yourself. |
| Search and monitoring | SpicrawlNoNo. Spicrawl fetches pages. | Crawl4AIStronger option: Cloud: search, monitorsThe cloud adds web search, direct answers with sources, ready-made recipes for some sites, and page monitors with webhooks. |
| AI agents | SpicrawlMCP, 25 tools + skillHosted MCP server with 25 tools, an agent skill, and one-command setup for Claude Code, Cursor and Codex. | Crawl4AIMCP, self-hosted or cloudAn MCP server on its self-hosted Docker server, with 7 tools, and a hosted one on the cloud for Claude, Cursor and other clients. |
| SDKs and languages | SpicrawlJS/TS SDK + CLIJavaScript and TypeScript SDK, a CLI, and HTTP examples in cURL, Python and TypeScript. | Crawl4AIPython library; RESTThe library is Python. Other languages use the Docker server’s REST API or the cloud API, with examples in cURL, Python and Node. |
| Open source | SpicrawlNoNo. Spicrawl is a hosted service. | Crawl4AIStronger option: Yes, Apache-2.0Yes. The library is Apache-2.0 on GitHub, with about 85,000 stars. The cloud is a paid hosted service. |
Switching from Crawl4AI
Crawl4AI’s quick start opens a browser on your machine and prints the page as Markdown. The Spicrawl request does the same with one POST and nothing to install: the response body is the Markdown itself. Ask for response_format=json when you want a JSON envelope instead.
import asynciofrom crawl4ai import AsyncWebCrawlerasync def main(): async with AsyncWebCrawler() as crawler: result = await crawler.arun("https://example.com") print(result.markdown)if __name__ == "__main__": asyncio.run(main())curl -X POST https://api.spicrawl.com/v1/scrape \ -H "Authorization: Bearer spicrawl_live_…" \ -H "Content-Type: application/json" \ -d '{ "url": "https://example.com", "response_format": "markdown" }'# Response: the Markdown itselfSpicrawl vs Crawl4AI: common questions
Is Spicrawl a good Crawl4AI alternative?
It is if you want a hosted API instead of running browsers yourself, and you scrape pages you already have URLs for. If you need open source, deep crawls that follow links, stealth mode or full control on your own servers, Crawl4AI is the better fit.
Is Crawl4AI free?
The library is free and open source under Apache-2.0, but you run and maintain the browsers, proxies and servers yourself. Crawl4AI Cloud is a paid hosted API with $10 of free credit to start, at launch pricing that may change. Spicrawl is free during its beta.
Is Spicrawl cheaper than Crawl4AI?
Spicrawl’s prices after the beta have not been announced, so a fair price comparison is not possible yet. The billing rules differ today: Spicrawl charges nothing for cache hits, while Crawl4AI Cloud charges half price for results from its archive. Both skip blocked and failed requests.
Can Spicrawl crawl a whole website like Crawl4AI?
No. Spicrawl does not discover URLs by following links. You send the pages you want, one at a time or up to 10,000 per batch call. Crawl4AI’s breadth-first, depth-first and best-first crawls do the discovery for you.
How do I switch from Crawl4AI to Spicrawl?
Replace crawler.arun(url) with a POST to https://api.spicrawl.com/v1/scrape, with your Spicrawl API key and response_format set to markdown. Spicrawl returns the Markdown as the response body. Deep crawls have no direct equivalent, so collect the URLs first and send them as a batch.
Sources
Crawl4AI details come from these pages, checked on 1 October 2026. Products change: if something here is out of date, email support@spicrawl.com and we will correct it.
- Crawl4AI on GitHub
- Crawl4AI Cloud and pricing
- Crawl4AI Cloud docs
- Crawl4AI docs: quick start
- Crawl4AI docs: deep crawling
- Crawl4AI docs: page interaction
- Crawl4AI docs: identity-based crawling
- Crawl4AI docs: proxy and security
- Crawl4AI docs: LLM extraction
- Crawl4AI docs: self-hosting and MCP
- Spicrawl docs: credits
Try Spicrawl free while it’s in beta
Run the same pages you scrape today and compare the results yourself. No credit card required.
Get free beta access