CaptchaAI Python SDK

Python client for the CaptchaAI captcha-solving API.

Call one method per captcha type; the SDK handles HTTP, polling, retries, and error mapping.

For full copy-paste examples of every captcha type and SDK feature, see examples.md.

Table of Contents

• Examples
• Installation
• Configuration
• Solve result
• Normal captcha
• Grid captcha
• BLS captcha
• reCAPTCHA v2
• reCAPTCHA v3
• Cloudflare Turnstile
• Cloudflare Challenge
• GeeTest
• CaptchaFox
• Friendly Captcha
• Lemin
• Other methods
• Error handling
• Proxies
• Async usage
• License

Installation

pip install captchaai

Configuration

Create a sync client with your API key. The constructor validates the key and reads the account thread cap via a startup threadsinfo call.

from captchaai import CaptchaAI

solver = CaptchaAI(
    api_key="YOUR_32_CHAR_API_KEY",
    proxy="user:pass@host:port",       # optional
    proxytype="HTTP",                  # required when proxy is set
    auto_retry=True,
    base_url="https://ocr.captchaai.com",
    thread_busy_timeout=120.0,
    max_retries=None,                  # optional; overrides auto_retry when set
)

A bad key raises InvalidKeyError before any solve runs. Call solver.close() when finished to release the HTTP client.

Solve result

Every solve method returns a SolveResult:

result.task_id     # str — task ID from submit (always present)
result.solution    # str | list | dict — token, cell indices, or structured answer
result.user_agent  # str | None — set for Enterprise / Cloudflare Challenge / CaptchaFox
result.raw         # dict | None — full API response (debugging)
result.raw_text    # str | None — unparsed solution string before JSON parsing

str(result)        # token string for common token types

Reuse result.user_agent on the target site when it is present (Enterprise, Cloudflare Challenge, CaptchaFox).

Normal captcha

Solve a text/image captcha. image may be a file path, URL, data-URI, raw base64, or bytes.

from captchaai import CaptchaAI

solver = CaptchaAI(api_key="YOUR_32_CHAR_API_KEY")

result = solver.normal(
    "captcha.png",
    numeric=1,              # 0=any, 1=digits, 2=letters, 3=digits+letters, 4=no digits
    min_len=4,
    max_len=6,
    phrase=0,               # 0=single word, 1=multi-word
    case_sensitive=0,       # 0=insensitive, 1=case-sensitive
    lang="en",
    instructions="Type the characters you see",
)
print(result.solution)

Grid captcha

Solve a tile-selection captcha. grid_size must be "3x3" or "4x4". solution is a list of cell indices.

result = solver.grid(
    "grid.png",
    instructions="select all traffic lights",
    grid_size="3x3",
)
print(result.solution)

BLS captcha

Solve a BLS multi-image captcha. Pass exactly 9 images (path/URL/base64/bytes each). solution is a list of cell indices.

result = solver.bls(
    images=[f"img{i}.png" for i in range(9)],
    instructions="YOUR_INSTRUCTIONS",
)
print(result.solution)

reCAPTCHA v2

Solve reCAPTCHA v2 (standard, invisible, or enterprise). Pass sitekey and page url. Enterprise results include user_agent.

# Standard
result = solver.recaptcha_v2(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
)
print(result.solution)

# Invisible
result = solver.recaptcha_v2(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
    invisible=True,
)

# Enterprise (optional action; result may include user_agent)
result = solver.recaptcha_v2(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
    enterprise=True,
    action="login",
)
print(result.solution, result.user_agent)

Optional keyword args: cookies, user_agent, proxy, proxytype.

reCAPTCHA v3

Solve reCAPTCHA v3 (standard or enterprise). The registry requires action for both variants.

result = solver.recaptcha_v3(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
    action="login",
    min_score=0.3,
)
print(result.solution)

# Enterprise
result = solver.recaptcha_v3(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
    action="login",
    enterprise=True,
)
print(result.solution, result.user_agent)

Optional keyword args: cookies, user_agent, proxy, proxytype.

Cloudflare Turnstile

Solve a Cloudflare Turnstile widget.

result = solver.turnstile(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
)
print(result.solution)

Optional keyword args: cookies, user_agent, proxy, proxytype.

Cloudflare Challenge

Solve a Cloudflare interstitial challenge page. A proxy is mandatory (client-level or per-call). The result includes user_agent to reuse on the target site.

solver = CaptchaAI(
    api_key="YOUR_32_CHAR_API_KEY",
    proxy="user:pass@host:port",
    proxytype="HTTP",
)

result = solver.cloudflare_challenge(url="https://example.com")
print(result.solution, result.user_agent)

Optional keyword args: cookies, user_agent, proxy, proxytype.

GeeTest

Solve GeeTest v3. solution is a dict with challenge, validate, and seccode.

result = solver.geetest(
    gt="YOUR_GT",
    challenge="YOUR_CHALLENGE",
    url="https://example.com",
)
print(result.solution["challenge"])
print(result.solution["validate"])
print(result.solution["seccode"])

Optional keyword args: cookies, user_agent, proxy, proxytype.

CaptchaFox

Solve a CaptchaFox slider challenge. A proxy is mandatory. Reuse result.user_agent when submitting the token.

solver = CaptchaAI(
    api_key="YOUR_32_CHAR_API_KEY",
    proxy="user:pass@host:port",
    proxytype="HTTP",
)

result = solver.captchafox(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
)
print(result.solution, result.user_agent)

Optional keyword args: cookies, user_agent, proxy, proxytype.

Friendly Captcha

Solve a Friendly Captcha proof-of-work challenge. No proxy is required.

result = solver.friendly_captcha(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
    version="v1",   # optional: "v1" or "v2"
)
print(result.solution)

Optional keyword args: proxy, proxytype.

Lemin

Solve a Lemin puzzle. solution is a dict with answer and challenge_uuid.

result = solver.lemin(
    captcha_id="YOUR_CAPTCHA_ID",
    div_id="lemin-cropped-captcha",
    url="https://example.com",
    api_server="api.leminnow.com",   # optional
)
print(result.solution["answer"])
print(result.solution["challenge_uuid"])

Optional keyword args: proxy, proxytype.

Other methods

Thread usage

CaptchaAI is thread-based. Check current usage:

info = solver.threads_info()
# {"threads": 10, "working_threads": 3}

Manual submit / fetch

Submit now and poll later (params use API names, e.g. googlekey, pageurl):

task_id = solver.send(
    "recaptcha_v2",
    googlekey="YOUR_SITEKEY",
    pageurl="https://example.com",
)

result = solver.get_result("recaptcha_v2", task_id)
print(result.solution)

Close

solver.close()

Error handling

All SDK errors inherit from CaptchaAIError:

from captchaai import (
    CaptchaAI,
    CaptchaAIError,
    InvalidKeyError,
    ValidationError,
    ProxyError,
    ThreadLimitError,
    NoThreadsError,
    UnsolvableError,
    APIError,
    NetworkError,
    TimeoutError,
)

solver = CaptchaAI(api_key="YOUR_32_CHAR_API_KEY")

try:
    result = solver.recaptcha_v2(
        sitekey="YOUR_SITEKEY",
        url="https://example.com",
    )
except InvalidKeyError:
    ...  # wrong-length key, or API rejected the key
except ValidationError:
    ...  # missing/invalid params, bad image input, missing required proxy, etc.
except ProxyError:
    ...  # bad proxy or proxy connection failed
except ThreadLimitError:
    ...  # all account threads busy (and retries skipped or exhausted)
except NoThreadsError:
    ...  # account expired or has no active plan
except UnsolvableError:
    ...  # service could not solve the captcha
except NetworkError:
    ...  # could not reach the API
except TimeoutError:
    ...  # poll deadline exceeded before a result was ready
except APIError:
    ...  # unexpected / malformed API response
except CaptchaAIError:
    ...  # any other SDK error

Retry behaviour

max_retries supersedes auto_retry when both are provided. Fatal errors (InvalidKeyError, ValidationError, UnsolvableError, NoThreadsError, and similar) always raise immediately.

Proxies

Proxies are supported at client level and per call. Format: "user:pass@host:port". proxytype is required whenever a proxy is set (HTTP, HTTPS, SOCKS4, or SOCKS5).

# Client-level (every solve)
solver = CaptchaAI(
    api_key="YOUR_32_CHAR_API_KEY",
    proxy="user:pass@host:port",
    proxytype="HTTP",
)

# Per-call override (does not mutate client config)
result = solver.recaptcha_v2(
    sitekey="YOUR_SITEKEY",
    url="https://example.com",
    proxy="user:pass@other-host:port",
    proxytype="SOCKS5",
)

Proxy is mandatory for:

• cloudflare_challenge
• captchafox

Async usage

Build the async client with await AsyncCaptchaAI.create(...) — not AsyncCaptchaAI(...) — because key validation must be awaited at construction.

import asyncio
from captchaai import AsyncCaptchaAI

async def main():
    solver = await AsyncCaptchaAI.create(api_key="YOUR_32_CHAR_API_KEY")

    result = await solver.turnstile(
        sitekey="YOUR_SITEKEY",
        url="https://example.com",
    )
    print(result.solution)

    await solver.aclose()

asyncio.run(main())

The async client mirrors the sync methods (normal, recaptcha_v2, send, get_result, threads_info, …). You can also use it as an async context manager:

async with await AsyncCaptchaAI.create(api_key="YOUR_32_CHAR_API_KEY") as solver:
    result = await solver.normal("captcha.png")
    print(result.solution)

Note: async with await AsyncCaptchaAI.create(...) requires Python 3.10+; alternatively assign the client first, then async with solver.

License

This project is licensed under the MIT License.

Copyright (c) 2026 Dev@Captchaai