Browser Crawl Engine
Crawling infrastructure

Rendered HTML from a real browser, by API.

Send up to 1,000 URLs in one request. We load each page in Chrome with JavaScript enabled, detect when a site blocks us, rotate the exit IP, retry, and hand you the HTML.

Create an account Read the quickstart

No card required to register. Accounts are reviewed before activation.

Python
import os

import requests

API = os.environ.get("API", "https://bankworks.info")
KEY = os.environ["API_KEY"]

# One request, twenty pages.
pages = range(1, 21)
urls = [f"https://example.com/p/{n}" for n in pages]

response = requests.post(
    f"{API}/v1/jobs",
    headers={"Authorization": f"Bearer {KEY}"},
    json={"urls": urls, "maxAttempts": 3},
    timeout=30,
)
print(response.json())
Node.js
const API = process.env.API ?? "https://bankworks.info";
const KEY = process.env.API_KEY;

// One request, twenty pages.
const urls = Array.from({ length: 20 }, (_, i) =>
  `https://example.com/p/${i + 1}`);

const response = await fetch(`${API}/v1/jobs`, {
  method: "POST",
  headers: {
    Authorization: `Bearer ${KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ urls, maxAttempts: 3 }),
});
console.log(await response.json());
cURL
# One request, twenty pages.
curl -sS -X POST "$API/v1/jobs" \
  -H "Authorization: Bearer $API_KEY" \
  -H 'Content-Type: application/json' \
  -d "$(jq -n '{
    urls: [range(1; 21) | "https://example.com/p/\(.)"],
    maxAttempts: 3
  }')"

What you actually get

This is infrastructure, not a dataset. You name the pages; we render them and return the HTML. What you extract from it is yours.

Real Chrome

Pages render in a current Chrome build with JavaScript on. Content assembled client-side is in the HTML you download, not missing from it.

Wait for what matters

Name a string only the real page contains and anything else is rejected and retried instead of stored. Scroll lazy-loaded content and wait after load before the HTML is read.

Blocks are detected

A challenge page that returns HTTP 200 is still a failure. We detect those, quarantine the exit IP and retry elsewhere instead of storing a useless page as a success.

1,000 URLs per call

Each URL becomes an independent task. One slow page does not hold up the other 999, and partial results are readable while the rest run.

Nothing is lost

Work is committed to a database before it is queued. A worker that dies mid-page has its task picked up again rather than dropped.

Verified downloads

Every HTML download carries an X-HTML-SHA256 header computed over the stored bytes, so you can prove what you received.

How it works

  1. Submit. POST /v1/jobs with your URLs and options. The response returns a jobId immediately — nothing blocks on rendering.
  2. We render. Each URL is claimed by a worker, loaded in Chrome through a managed exit IP, and checked for block signals before the result is accepted.
  3. Poll or wait. GET /v1/jobs/{jobId} returns counters and per-task status.
  4. Download. GET /v1/tasks/{taskId}/html returns the stored HTML with its SHA-256.

Full quickstart, with a working script →

Honest limits

Numbers we measured on our own cluster, not marketing figures.

12.8s Median full page render Measured on a 1.25 MB listing page.
1,000 URLs per request Hard limit, enforced at validation.
10/s Default API rate limit Burst 20. Raised per account on request.
10 Attempts per URL, at most Default 3. A blocked page retries on a different exit.

Throughput is shared across the fleet. If you need a committed rate, tell us the volume and we will quote capacity rather than guess.

What we will not do

We do not break anti-bot protection. No CAPTCHA solving, no browser fingerprint spoofing, no residential-proxy evasion. Sites that actively refuse automated access will keep refusing it.

We also do not resell consumer bandwidth. Every exit is one we operate under contract.

You are responsible for having the right to collect the pages you ask for. We respond to complaints by blocking targets.

Monthly plans

Prefer a predictable bill? A monthly plan renews by itself through Stripe or PayPal and includes a number of successful requests. You can also keep a balance next to it: usage beyond your plan is charged from the balance at 1.45 USD per 1,000 requests.

Starter

100,000 successful requests every month.

140.00 USD / month

  • 100,000 successful requests a month
  • 1.40 USD per 1,000 requests
  • Renews automatically, cancel any time
Get started

Growth

150,000 successful requests every month.

200.00 USD / month

  • 150,000 successful requests a month
  • 1.33 USD per 1,000 requests
  • Renews automatically, cancel any time
Get started

Scale

200,000 successful requests every month.

250.00 USD / month

  • 200,000 successful requests a month
  • 1.25 USD per 1,000 requests
  • Renews automatically, cancel any time
Get started

Cancel any time: you keep what you paid for until the end of the month. Card payments are verified with your bank (3-D Secure). Failed and blocked requests are never billed.

Start with the free tier

Register, get reviewed, and receive an API key. Pay per successful request from a balance, or pick a monthly plan — a request that did not succeed is not billed.

Create an account See pricing