capzy — official Python SDK

Solve every major captcha with a single line of Python.

Website · Dashboard · Docs · Pricing · Get free credits

Why Capzy

Built by people who were tired of paying half a cent to get back errorCode: ERROR_NO_AVAILABLE_WORKERS.

• ⚡️ Fast. Median solve <5s for Turnstile / reCAPTCHA v2 / hCaptcha.
• 💸 Cheap. Token-bound captchas start at $0.40 per 1,000 solves.
• 🧠 Solves what others don't. Native support for Temu, Shopee (slider + curve), Binance bCAPTCHA2, CaptchaFox, MTCaptcha, Lemin, ALTCHA, Yidun, Capy — not just the obvious ones. (hCaptcha is in active development and not currently sold; status: see roadmap.)
• 🧩 Drop-in compatible. Uses the same createTask / getTaskResult shape your existing scripts already speak. Migration is a one-line base_url change.
• 🆓 Free credits on sign-up. Every new account gets $0.10 in real credits to test in production. No card required.
• 🔓 No retainer, no minimum. Pay-as-you-go. Top up when you want.
• 📊 Real dashboard. Per-key analytics, score quality charts for v3, full payment audit log.

Create a free account → Get $0.10 in free credits the moment you sign up — enough to verify integration end-to-end on roughly 80 reCAPTCHA v3 or 80 Turnstile solves before you pay a cent.

Install

pip install capzy

Requires Python ≥ 3.9 and requests. That's it.

Bleeding-edge (straight from main):

pip install "git+https://github.com/capzy-ai/capzy-pip.git"

60-second quickstart

from capzy import CapzyClient

capzy = CapzyClient("capzy_xxxxxxxxxxxxxxxxxxxxxxxx")   # from capzy.ai/dashboard

solution = capzy.solve(
    type="AntiTurnstileTaskProxyLess",
    website_url="https://example.com/login",
    website_key="0x4AAAAAAA...",
)
print(solution["token"])     # paste into the cf-turnstile-response field

That's the whole pattern. solve() does createTask → polls getTaskResult → hands you the solution dict.

Need to route the solve through your own proxy? Switch to the …Task (non-ProxyLess) type and add the proxy params:

solution = capzy.solve(
    type="AntiTurnstileTask",
    website_url="https://example.com/login",
    website_key="0x4AAAAAAA...",
    proxy_type="http",
    proxy_address="123.45.67.89",
    proxy_port=8080,
    proxy_login="user",
    proxy_password="pass",
)

Need lower-level control? Use the raw verbs:

created = capzy.create_task(
    type="AntiTurnstileTaskProxyLess",
    website_url="https://example.com/login",
    website_key="0x4AAAAAAA...",
)
task_id = created["taskId"]

# Poll on your own schedule:
status = capzy.get_task_result(task_id)

Param naming. Pass either website_url= / website_key= (Python style) or websiteURL= / websiteKey= (wire style). They're equivalent — the SDK normalises snake_case to the camelCase the API expects.

How it works

┌──────────────┐  createTask   ┌──────────┐    enqueue    ┌──────────┐
│ your script  │ ────────────▶ │  Capzy   │ ────────────▶ │ solver   │
│  (capzy SDK) │ ◀──────────── │   API    │ ◀──────────── │  cluster │
└──────────────┘ getTaskResult └──────────┘   solution    └──────────┘
                          (polls until "ready")

Submit → poll → done. No webhooks to set up, no SDK runtime required on the solver side, no captive browser to manage.

Supported captchas

Every captcha below is supported as a proxyless task (we handle the upstream routing) and a proxy task (you supply the upstream IP). All prices are USD per solve.

Token-based — the workhorses

Anti-bot platforms

Image-based

Full live pricing at capzy.ai/pricing — these tables are scraped from the same source-of-truth.

Examples per service

Cloudflare Turnstile

from capzy import CapzyClient

capzy = CapzyClient("capzy_xxx")
sol = capzy.solve(
    type="AntiTurnstileTaskProxyLess",
    website_url="https://example.com",
    website_key="0x4AAAAAAA...",
    action="login",      # optional — must match data-action
    cdata="user-123",    # optional — server-issued cdata
)
print(sol["token"])

reCAPTCHA v2

from capzy import CapzyClient

capzy = CapzyClient("capzy_xxx")
sol = capzy.solve(
    type="ReCaptchaV2TaskProxyLess",
    website_url="https://example.com",
    website_key="6Lc_aCMTAAAAAB...",
    is_invisible=False,
)
print(sol["gRecaptchaResponse"])

reCAPTCHA v3 (score-based)

from capzy import CapzyClient

capzy = CapzyClient("capzy_xxx")
created = capzy.create_task(
    type="ReCaptchaV3TaskProxyLess",
    website_url="https://example.com",
    website_key="6Lc_aCMTAAAAAB...",
    page_action="checkout",
    min_score=0.7,
)
task_id = created["taskId"]

# After you call Google siteverify with your secret, report the score back —
# it lights up your v3 quality dashboard and helps us auto-route your traffic.
capzy.report_score(task_id, score=0.9, action="checkout", hostname="example.com")

GeeTest v4

from capzy import CapzyClient

capzy = CapzyClient("capzy_xxx")
sol = capzy.solve(
    type="GeeTestV4TaskProxyLess",
    website_url="https://example.com",
    captcha_id="abc123...",
)
# sol contains captcha_id, lot_number, pass_token, gen_time, captcha_output

Image-to-Text (sync solve — no polling)

import base64
from capzy import CapzyClient

capzy = CapzyClient("capzy_xxx")
with open("captcha.png", "rb") as f:
    body = base64.b64encode(f.read()).decode()

sol = capzy.solve(type="ImageToTextTask", body=body, case="lower")
print(sol["text"])

More copy-pasteable runners in examples/.

Free credits & getting an API key

1. Create an account at capzy.ai/auth/register.
2. We post a $0.10 welcome bonus as a real credit transaction on your dashboard. (Audit it like any other payment.)
3. Generate an API key on capzy.ai/dashboard and drop it into CapzyClient("capzy_…").

The $0.10 is plenty to verify end-to-end:

When you run out, top up with Stripe (card / Apple Pay / Google Pay) or MixPay (USDT / BTC / 30+ coins). No subscription, no minimum.

Handling errors

from capzy import CapzyClient, ApiError, TaskFailedError, TaskTimeoutError

capzy = CapzyClient("capzy_xxx")
try:
    sol = capzy.solve(
        type="AntiTurnstileTaskProxyLess",
        website_url="https://example.com",
        website_key="0x4AAAAAAA...",
    )
except ApiError as e:
    # createTask was rejected — bad key, wrong task type, insufficient funds, etc.
    print("API rejected the task:", e.error_code, e.error_description)
    if e.recommended_task_type:
        print("Hint — submit as:", e.recommended_task_type)
except TaskFailedError as e:
    # Task ran but couldn't be solved. Refunded automatically.
    print("Solver failed:", e.error_code, e.error_description)
except TaskTimeoutError as e:
    # We didn't get an answer within max_wait. Refunded automatically.
    print(f"No answer after {e.waited:.0f}s")

You are never charged for failed or timed-out tasks. Refunds post as real credit transactions, visible in your dashboard.

Configuration

capzy = CapzyClient(
    api_key="capzy_xxx",
    base_url="https://api.capzy.ai",   # override for on-prem / staging
    timeout=30.0,                       # per HTTP-request timeout
)

# Tune the polling behaviour per .solve() call:
capzy.solve(type="AntiTurnstileTaskProxyLess", website_url=..., website_key=...,
            poll_interval=1.0, max_wait=90.0)

Need a proxy on the SDK's own egress, retries, or custom TLS? Pass your own requests.Session:

import requests
from capzy import CapzyClient

session = requests.Session()
session.proxies = {"https": "http://user:pass@proxy.example:8080"}
capzy = CapzyClient("capzy_xxx", session=session)

API reference

CapzyClient is a thin wrapper around the Capzy HTTP API:

All task parameters travel as kwargs. Snake_case (website_url, proxy_address, min_score) and camelCase (websiteURL, proxyAddress, minScore) are equivalent — the SDK normalises to the wire format.

FAQ

Do you need to supply a proxy? No — every *ProxyLess task type runs against our own network. Bring your own only when the target binds the token to your egress IP (some anti-bot platforms do).

Do you support async? Not in v0.0.1. The sync client is intentionally tiny — pair it with a thread pool, anyio.to_thread, or asyncio.to_thread if you need fan-out today. A native AsyncCapzyClient is on the roadmap.

What happens if a task fails? The credit is automatically refunded as a refund transaction. The SDK raises TaskFailedError so you can retry with a different task type or proxy.

Can I run this in a long-running service? Yes. Re-use one CapzyClient across threads — the underlying requests.Session is thread-safe for our usage pattern.

Where do I find my API key? capzy.ai/dashboard → API keys → New key.

Do you have rate limits? Per-key concurrency limits exist (visible on your dashboard). When you hit them, createTask returns ERROR_RATE_LIMIT and the SDK raises ApiError(error_code="ERROR_RATE_LIMIT") so you can back off.

Where do solutions need to be submitted?

• Turnstile → cf-turnstile-response form field
• reCAPTCHA v2 / Enterprise → g-recaptcha-response
• Others → service-specific; see the per-service guides at capzy.ai/docs

Roadmap

Captchas not yet on the price list but actively being worked on:

• hCaptcha (incl. Enterprise) — local CV pipeline in late-stage testing; SDK task types (HCaptchaTaskProxyLess, etc.) already accepted so your code is forward-compatible.
• Native AsyncCapzyClient — asyncio-native client, same API surface.
• Streaming solveMany() — fan out N tasks, yield as they finish.

Want something prioritised? Open a GitHub issue or email support@capzy.ai.

Links

• 🌐 Website: https://capzy.ai
• 📚 Documentation: https://capzy.ai/docs
• 💵 Pricing: https://capzy.ai/pricing
• 🔑 Dashboard / API keys: https://capzy.ai/dashboard
• 🆓 Free credits on sign-up: https://capzy.ai/auth/register
• 📜 Changelog: CHANGELOG.md
• 🐛 Issues: https://github.com/capzy-ai/capzy-pip/issues
• 💬 Support: support@capzy.ai

License

MIT — see LICENSE. Use it in proprietary software, redistribute it, fork it. Just don't blame us if your bot finds love.