Updated November 2026 · 5 min read
You're on a deadline. You write requests.get(url), and you get back a 403 with a "Checking your browser" page. You Google "Cloudflare bypass" and get buried in Selenium tutorials that require you to install Chromium on a server that has 2 GB of RAM. There's a simpler way.
Cloudflare checks four things on every request:
requests uses OpenSSL with a signature that differs from Chrome's.You can't fake all four from plain Python. That's why every "pure requests Cloudflare bypass" library on GitHub is dead within 3 months — Cloudflare patches it.
Use a bypass API as a drop-in replacement for the origin. It handles all four checks for you and returns raw HTML exactly as the origin sent it.
import requests
resp = requests.post(
"https://cloudflare-bypass.shop/v1/solve",
headers={"X-API-Key": "YOUR_KEY"},
json={"url": "https://target.com"},
timeout=200
)
print(resp.status_code, len(resp.text))
That's it. resp.text is the raw page. If the target is an image or PDF, resp.content gives you bytes. The Content-Type header matches the origin. No JSON wrapper, no base64 encoding.
import requests
API = "https://cloudflare-bypass.shop/v1/solve"
KEY = "YOUR_KEY"
def fetch(url):
r = requests.post(
API,
headers={"X-API-Key": KEY},
json={"url": url},
timeout=200
)
if r.status_code == 200 and len(r.text) > 10000:
return r.text
raise RuntimeError(f"Failed: {r.status_code}, len={len(r.text)}")
html = fetch("https://www.example.com/protected-page")
print(html[:500])
Here's the single most useful rule in scraping: if the response is 200 OK and the body is longer than 10,000 characters, you almost certainly got the real page. Cloudflare's interstitial is a few hundred bytes. Empty pages are a few hundred. Real pages are tens of thousands. Stop debugging, start parsing.
A good bypass API distinguishes between:
CF Bypass follows exactly this model: $0.03 per success, $0.01 per 404, refund on failure. So you never pay for a page that didn't exist.
Sometimes you need the page after JavaScript runs — for infinite scroll, clicking buttons, or executing site JS. That's a different problem. Bypass APIs usually offer a JS rendering option, or you can run Playwright with a residential proxy. But for 90% of scraping tasks, fetching raw HTML is enough — most sites still render useful content server-side.
| Error | Cause | Fix |
|---|---|---|
| 403 from API | Invalid or expired API key | Regenerate key, check balance |
| 402 / no balance | Wallet not topped up | Send USDT/USDC, wait 2 min |
| Timeout | Target hung or Cloudflare stuck | Retry once; if again, target is down |
| Small body | Target returned a real short page | Check status, not length |
Not directly. Cloudflare fingerprints your TLS handshake, HTTP/2 settings and IP reputation, and the standard requests library fails all three. You need a bypass API that handles these on its side and returns raw HTML.
No, for most scraping tasks you don't. Selenium and Playwright are only needed when the page requires interaction after JavaScript runs. For fetching raw HTML, a bypass API is faster and cheaper.
CF Bypass charges $0.03 per successful bypass, $0.01 for a 404, and refunds failures. You never pay for a page that did not load.
Check for a 200 OK status, a body longer than 10,000 characters, and a matching Content-Type. If all three match, you got the real page. Do not search the HTML for the word "cloudflare" — every protected page contains it.
Related: Python guide · ZenRows alternatives · Scrapy + Cloudflare · API comparison 2026