Cloudflare · DataDome · Akamai · PerimeterX
Any URL to clean Markdown — protected pages included
For LLM pipelines, AI agents and data teams. Plain pages, JavaScript apps and pages behind Cloudflare, DataDome, Akamai or PerimeterX, as Markdown or HTML — at $0.50 per 1,000 pages, with no credit multipliers for rendering or bot protection.
- ▸Protected pages at the price of plain ones. A page behind Cloudflare, DataDome, Akamai or PerimeterX costs the same $0.50 / 1k as example.com.
- ▸Own exits, real browsers. A mobile proxy pool and Chrome fleet we run ourselves — not a resold unblocker. (See who runs it.)
- ▸Pay only for delivered pages. Blocks, timeouts and 404s are never billed. Credit never expires.
Integration
One request. The page as Markdown or HTML.
No headless fleet to babysit, no proxy rotation, no challenge solver. Send a URL, get the content — or just the bare document with Accept: text/markdown.
curl -X POST https://reader.hilgard.cz/v1/read \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"url": "https://shop.example/products/espresso-grinder"}'
# or just the Markdown:
curl "https://reader.hilgard.cz/v1/read?url=https://example.com" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Accept: text/markdown"{ "url": "https://shop.example/products/espresso-grinder", "final_url": "https://shop.example/products/espresso-grinder", "status": 200, "title": "Espresso Grinder EG-1 | Shop", "format": "markdown", "content": "# Espresso Grinder EG-1\n\n| Spec | Value |\n| --- | --- |\n| Burrs | 64 mm flat |\n| Price | $349 |\n\n## Reviews\n…", "length": 18342, "total_length": 18342, "truncated": false, "next_index": null, "fetch": { "mode": "challenge", "challenge": "cloudflare", "from_cache": false, "ms": 14210 }, "billed": true }
| Parameter | Type | Default | Description |
|---|---|---|---|
| url | string | required | The page to read: a public http or https URL (ports 80 and 443). |
| format | string | "markdown" | "markdown" or "html". |
| only_main_content | boolean | true | Drop navigation, headers, footers and cookie bars. false = the whole rendered page. |
| include_links | boolean | true | Keep links as absolute URLs; false for prose only (fewer tokens). |
| max_length | number | 100000 | Characters per call, 1,000–100,000. Longer documents come in slices. |
| start_index | number | 0 | Offset of the next slice (next_index of the previous answer). Slices within 15 minutes are free. |
Full API reference → · Auth: Authorization: Bearer YOUR_API_KEY · the same key works for the SERP API.
Search → read, one key
Find the pages on Google, then read them.
Your key also runs the Google SERP API: organic results and the AI Overview with every cited source. Read the pages that rank or get cited — for grounding an LLM, a GEO report or a content gap analysis — without a second account.
- ▸One key, one credit. Searches and page reads come out of the same prepaid balance.
- ▸Real URLs.
resolve_urls: truereturns the pages' own addresses, ready to read. - ▸$3.70 per 1,000 researched queries on Starter: a search with its AI Overview ($1.20 / 1k) plus five pages read ($0.50 / 1k each).
import requests
H = {"Authorization": "Bearer YOUR_API_KEY"} # one key for both
# 1. Search: rankings and the AI Overview with its cited sources
serp = requests.post("https://serp.hilgard.cz/v1/search", headers=H, timeout=180, json={
"query": "best crm for startups",
"country": "us",
"resolve_urls": True, # real page URLs instead of Google redirect links
}).json()
sources = (serp.get("ai_overview") or {}).get("sources", [])
urls = [s["url"] for s in sources][:5] or [r["url"] for r in serp["organic_results"]][:5]
# 2. Read: every page as Markdown, protected pages included
for url in urls:
page = requests.post("https://reader.hilgard.cz/v1/read", headers=H, timeout=180,
json={"url": url}).json()
print(page.get("final_url"), page.get("fetch", {}).get("mode"), len(page.get("content", "")))How a page is read
The cheapest way that works, picked for you.
A browser-like request through our own exits. Most pages are done here in 1–3 seconds.
An unrendered React, Vue or Angular shell is detected and the page is rendered in a headless browser.
A bot check is retried on another exit first, then passed in a real Chrome on a mobile IP — about 10–30 seconds. Akamai and PerimeterX sites mostly pass on our mobile exits without one.
Whichever way the page was read, it costs the same $0.50 per 1,000. Failed reads cost nothing.
Pricing
One price per page. No multipliers.
JavaScript rendering, Cloudflare, DataDome, Akamai, PerimeterX, Markdown or HTML — the price does not change. Price per 1,000 delivered pages; prices exclude VAT and sales tax, which are added at checkout where they apply.
Starter
up to 1M pages / month
$0.50 / 1k pages
Prepaid credit, self-service key, e-mail support
Growth
1M – 10M pages / month
$0.35 / 1k pages
By agreement: custom limits, Slack support
Scale
10M+ pages / month
$0.25 / 1k pages
By agreement: dedicated capacity, custom limits
- ▸ 5,000 free pages for every new account, no credit card.
- ▸ Only delivered pages (HTTP 200) are billed; a long document once.
- ▸ Prepaid credit that never expires — no subscription, nothing lapses.
- ▸ One credit and one key for the reader and the SERP API.

Who's behind it
Built by Jan Hilgard — building infrastructure since 2002.
I founded the hosting company Hosting90 in 2002 and ran it for eighteen years, until its sale to WY Group in 2020. Today I build self-healing data infrastructure: the reader runs on the same mobile proxy pool and browser fleet as my Google SERP API, built and operated in-house.
No ticket queue: you talk directly to the engineer who runs the proxies, the browsers and the extraction.
FAQ
Reader API questions, answered.
What does the reader API do?
You send a URL and get the page back as clean Markdown or HTML, ready for an LLM, a RAG pipeline or your own parser. Plain pages are fetched over HTTP through our mobile exits, JavaScript apps are rendered in a browser, and pages behind Cloudflare or DataDome bot checks are read in a real Chrome — you do not choose, the API picks the cheapest way that works.
Which bot protections does it get past?
Cloudflare (including managed challenges and Turnstile) and DataDome (including its device check) are passed in a real Chrome on a mobile IP. Most sites behind Akamai Bot Manager and PerimeterX / HUMAN are read on our mobile exits without a challenge at all. Kasada-protected sites mostly fail today, and no service gets through every site every time: a failed read is not billed, and fetch.mode and fetch.challenge tell you how each page was read. Interactive CAPTCHAs that ask a person to solve a puzzle are not solved.
Do protected pages cost more?
No. A page behind Cloudflare, DataDome, Akamai or PerimeterX costs the same as example.com: $0.50 per 1,000 delivered pages on prepaid credit, with no credit multipliers for JavaScript rendering, bot protection or the output format. The response tells you how the page was read (fetch.mode: http, browser or challenge).
What do I pay for failed reads?
Nothing. Only a page delivered with HTTP 200 is billed. Blocks, timeouts, pages that answer 404 and invalid URLs are free. A long document is billed once: further slices of it (start_index) within 15 minutes are free.
Markdown or HTML?
Set format to "markdown" (default) or "html". With only_main_content (default true) navigation, headers, footers and cookie bars are removed and you get the article or product content; set it to false for the whole rendered page. Links and images are absolute URLs; tables come out as GitHub-flavored Markdown.
Is it the same key as serp.hilgard.cz?
Yes. One account and one API key work for both the SERP API (search Google, AI Overviews) and the reader (read the pages it returns). Free pages and free searches are counted separately; paid usage of both comes out of the same prepaid credit, which never expires.
How fast is a read?
A plain page usually comes back in 1–3 seconds, a JavaScript page in 5–15 seconds, and a page behind a bot challenge in about 10–25 seconds, because a real browser has to pass the check first. Repeated reads of the same page within 15 minutes come from cache.
What may I read with it?
Public web pages. The reader does not log in anywhere and must not be used for personal data or content behind a login; you are responsible for respecting the terms of the sites you read. Private and internal addresses are refused. See the terms of use.
Throw your hardest URLs at it.
Enter your e-mail and get an api_key with 5,000 free pages — the same key also runs the SERP API.
Instant API key delivery · No credit card required