hcaptcha-challenger vs a solving service.

hcaptcha-challenger is one of the most-starred open-source hCaptcha solvers on GitHub: a Python library that works the challenge inside a Playwright browser using Google's Gemini models. NoneCap is an hCaptcha solver service that returns the token from an API call. This page sets out what hcaptcha-challenger does and needs, quoted from its README, docs, releases and issue tracker as of November 2, 2026, and when each one is the better pick. One disclosure: this site is NoneCap.

What hcaptcha-challenger does

Its GitHub repository describes it in one line: "Gracefully face hCaptcha challenge with multimodal large language model." The README adds that it "does not use any third-party anti-captcha services". Its documentation says the agent "controls browser pages through Playwright" and "interacts exclusively with web pages through the Page object". In practice:

• You open the page with Playwright, in any browser Playwright can launch, and pass the Page to its agent. The agent clicks the hCaptcha checkbox and works through the challenge in that page.
• Each step asks a Gemini model through your own key. Its main branch moved the default to Gemini 3.7 Flash in August 2026.
• Its README lists the challenge types the agent handles: binary image labels, point selection and drag-drop. Bounding-box and multiple-choice challenges are listed without a check mark.
• It is free under GPL-3.0. On November 2, 2026 it had about 2,600 stars and 450 forks on GitHub, and pypistats.org counted about 26,000 downloads in the previous month.

What it needs from you

• Python 3.10 or newer, per its PyPI page. Install with uv pip install hcaptcha-challenger.
• Playwright and a browser. The agent only works through a Playwright page, so every solve runs a browser on your machine or server.
• A Gemini API key, set as GEMINI_API_KEY. Its configuration stops with "GEMINI_API_KEY is required but not provided" without one, and Google bills the model calls to that key. An open request to add self-hosted, OpenAI-compatible models (#1098, June 2026) says the code "currently hardcodes Gemini".
• Time to keep up with hCaptcha. Its open issues include requests for challenge types it does not handle yet, such as "Drag the pipe from the right to complete the puzzle", and a January 2026 report of an infinite retry loop on one challenge.

Maintenance as of November 2026

The latest release is v0.19.0, uploaded to PyPI on December 21, 2025 and tagged on GitHub releases on January 5, 2026, with the note "Gemini 3". Since then the main branch has had three commits, the latest on August 15, 2026 (PR #1101, the new default model). The repository is not archived. Several 2026 issues had no maintainer reply on November 2, among them a report that the older local YOLO and ONNX models can no longer be selected in v0.19 (#1091). A question about Enterprise rqdata sitekeys (#1077, October 2025) is also unanswered.

hcaptcha-challenger vs NoneCap

When self-hosting is the better choice

Pick hcaptcha-challenger, or another open-source project, when:

• you want to read and change the code, or study how a multimodal model handles hCaptcha;
• you already run Playwright and want the solving step inside your own Python process, with no solving vendor in between;
• you need a few solves now and then, and Gemini usage on your own key costs you less than paying per solve;
• you are building your own solver and want its dataset tool (hc dataset collect).

The price of that freedom is upkeep. When hCaptcha adds a challenge type or a dependency changes, you wait for a fix or write it yourself.

Where a solving service fits

NoneCap fits when you would rather not maintain the solving step. Send the sitekey and page URL and you get an hCaptcha token back for your code to submit with the form, from Playwright, requests, any other HTTP client or any language. You need no model key and no browser for the challenge, and regular, invisible and Enterprise rqdata sitekeys are covered. Failed solves are never charged.

import os, time, requests

API = "https://api.nonecap.com/v1/solves"
HEADERS = {"Authorization": f"Bearer {os.environ['NONECAP_KEY']}"}

def solve_hcaptcha(sitekey: str, url: str) -> str:
    r = requests.post(API, headers=HEADERS, params={"wait": 90},
                      json={"type": "hcaptcha", "sitekey": sitekey, "url": url},
                      timeout=120)
    r.raise_for_status()
    data = r.json()
    while data["status"] in ("pending", "solving"):
        time.sleep(2)
        data = requests.get(f"{API}/{data['id']}", headers=HEADERS, timeout=30).json()
    if data["status"] != "solved":
        raise RuntimeError(f"solve {data['status']}: {data.get('error')}")
    return data["token"]

With ?wait=90 the request stays open for up to 90 seconds and the token usually comes back in that response; the loop covers the rest. The Playwright guide shows where the token goes on the page, and the API reference has examples in cURL, Node, Python and Go. Pricing is $0.20 to $0.50 per 1,000 credits, from 1 credit per challenge round; new accounts get 1,300 credits with no card. For hCaptcha in your own Chrome, the NoneCap extension gives 100 free solves a day with no account.

Which one to use

If you want open source you can read and change, and you are happy to run a browser and a Gemini key and keep up with hCaptcha's changes, hcaptcha-challenger is the project to start from. If you want a token from an API call, with Enterprise sitekeys covered and failed solves not charged, NoneCap's API does that.