Skip to content
reCAPTCHA v2 Solver

How to Solve reCAPTCHA 2: Automatic Bypass Guide

A complete guide on how to solve reCAPTCHA 2 automatically by image click recognition: 3x3 and 4x4 grids, dynamic tiles, the API flow and integration with Selenium, Puppeteer and Playwright.

Start solving

What Is reCAPTCHA v2 ​

Before looking at how to solve reCAPTCHA 2, it helps to understand what this protection does. reCAPTCHA v2 is Google's widely deployed verification system that tells people and automated scripts apart. It appears on millions of websites as the familiar "I'm not a robot" checkbox. When a user clicks the checkbox, Google evaluates behavioral signals such as mouse movement, click timing, the browser and cookie history. If the risk looks low, the user passes right away. Otherwise, an image selection challenge appears.

The image challenge shows a grid of photos and asks the user to select all tiles with a specific object, such as traffic lights, crosswalks, buses or bicycles. Google keeps rotating these images and refining its detection with machine learning, which makes reCAPTCHA v2 one of the most difficult captchas to bypass with simple automation scripts. This is why developers and automation engineers need a reliable way to solve reCAPTCHA 2 at scale.

You can try reCAPTCHA v2 on the reCAPTCHA demo page.

☑️

Checkbox Challenge

The classic "I'm not a robot" widget. Google evaluates behavioral signals when the user clicks the checkbox.

🖼️

Image Grid Challenge

A 3x3 or 4x4 grid of images. The user must select all tiles with the requested object to continue.

🕶️

Invisible reCAPTCHA v2

No visible checkbox. The check runs in the background and shows an image challenge only when it is in doubt.

How to Solve reCAPTCHA 2 via Image Click Recognition ​

In a browser, the most practical way to solve reCAPTCHA 2 is image click recognition. Your script captures the full image grid, sends it with the task instruction to the solver API and gets back the exact coordinates of the tiles to click. This works with all reCAPTCHA v2 image grids, including 3x3 and 4x4 grids with any of the supported objects.

Your script needs the complete captcha image as the user sees it and the instruction text that says which objects to select (for example, "Select all images with bus" or "Select all images with traffic light"). The image can be sent as a file (multipart) or in Base64.

If you do not need a browser at all and just want a ready g-recaptcha-response token, use reCAPTCHA v2 token solving instead.

💡 Supported reCAPTCHA v2 tasks include cars, bicycles, boats, bridges, buses, chimneys, crosswalks, hydrants, motorcycles, mountains, palm trees, traffic lights, tractors, taxis, stairs and parking meters. Other tasks are supported as well, and new ones are added regularly.

Step by Step: How to Solve reCAPTCHA 2 Automatically ​

Below is the complete flow for solving reCAPTCHA v2 by image click recognition with the Cap.Guru API. It fits web scraping, testing and other automation pipelines.

Capture the Image and Task Text

Take one screenshot of the full reCAPTCHA v2 image grid and note the instruction text above it (for example, "Select all images with crosswalks"). The image can be sent as a file (multipart) or in Base64.

Submit the Task to the API

Send a POST request to http://api2.cap.guru/in.php with key (your API key), method (post or base64), textinstructions (the task text, for example crosswalks), click (recap for coordinates or recap2 for cell numbers) and file or body (the image). The server returns a task ID.

Get the Solution

Wait 5 seconds and send a GET request to http://api2.cap.guru/res.php with your key, action=get and the task ID. With click=recap, the answer looks like coordinate:x=44,y=32;x=143,y=11, where x=0,y=0 is the upper left corner of the image. With click=recap2, the answer is cell numbers: 1,2,6,9. If the result is not ready, you get CAPCHA_NOT_READY; repeat the request after 5 seconds.

Click the Coordinates

Click the returned coordinates in the image grid one by one. With cell numbers (recap2), click the matching cells. If reCAPTCHA shows another grid, repeat the process.

Parameters and code examples are in the reCAPTCHA v2 image click API documentation.

Solving reCAPTCHA 2 with Different Automation Tools ​

Knowing how to solve reCAPTCHA 2 is half the job, the other half is connecting it to your automation stack. In Selenium, find the captcha image element, take a screenshot, send it to the API and click the returned coordinates with ActionChains. In Puppeteer and Playwright, capture the grid with page.screenshot() and click with page.mouse.click() at the returned positions.

If you work in Python with Playwright or Selenium, the Python module does all of this with a single solve_recaptcha2() call, and the Node.js module does the same in Puppeteer.

The click based approach is particularly effective because it closely mimics real human interaction with the captcha. Instead of injecting tokens into hidden form fields, your script physically clicks on the correct grid cells, which makes the solution more resistant to detection by Google's behavioral analysis system. The API supports two response formats: coordinates (recap) for exact pixel positions and cell numbers (recap2) for grid-based clicking. Choose the one that fits your code.

reCAPTCHA v2 Image Grid Versions ​

reCAPTCHA uses two main grid layouts: a 3x3 grid (9 cells) and a 4x4 grid (16 cells). Both show a composite image divided into tiles, with an instruction to select all tiles with a specific object. The Cap.Guru API supports both and returns coordinates or cell numbers accordingly.

The challenge may also use dynamic grids, where tiles are replaced with new images after you click them. In that case, take a fresh screenshot after each round of clicks and send it again.

📋

3x3 Grid (9 cells)

The standard layout with 9 tiles. Use click=recap for pixel coordinates or click=recap2 for cell numbers 1 to 9.

🔍

4x4 Grid (16 cells)

The extended layout with 16 tiles. The same parameters apply, and the coordinates match the larger grid.

🔄

Dynamic Replacement

Tiles refresh after a click. Take a new screenshot, send it again and click the new coordinates until the challenge is complete.

Common Errors When Solving reCAPTCHA 2 via Clicks ​

Even with a reliable captcha solver, several pitfalls can prevent a successful reCAPTCHA v2 bypass. The most frequent issue is a cropped or low quality image. The API needs the complete grid as one image; partial screenshots or images with browser UI around them give wrong results. Another common mistake is the wrong task text: textinstructions must match the instruction in the captcha (for example, bus when the challenge asks for buses).

Coordinate offsets are another cause of missed clicks. If the screenshot size differs from the size of the captcha on the page, the clicks land in the wrong place. Keep them the same. Tasks are supported in English only, so always send the instruction in English, whatever the page language. If reCAPTCHA shows another grid after your clicks, take a new screenshot and send a new task for each round.

💡 If your clicks keep missing, check three things: image quality (the full grid, no cropping), coordinate alignment (the screenshot size matches the displayed size) and the task text (in English and matching the captcha instruction).

Best Practices for Automated reCAPTCHA 2 Solving ​

To keep a high success rate when solving reCAPTCHA 2 at scale, follow these guidelines. Use high quality residential or mobile proxies and rotate them regularly; datacenter IPs are more likely to trigger hard image challenges or outright blocks. Add realistic delays between your clicks to simulate natural human interaction. Always use up to date browser fingerprints and User-Agent strings, as Google flags requests from outdated or headless browser signatures.

Implement proper error handling and retry logic: if the API returns CAPCHA_NOT_READY, wait 5 seconds before polling again. For dynamic grids, take a new screenshot and send a new task for each round until the captcha is passed. The Cap.Guru API accepts tasks in parallel, so you can queue many reCAPTCHA v2 image tasks at once and process the results as they arrive.

Why Cap.Guru Is the Fastest Way to Solve reCAPTCHA 2 ​

Cap.Guru provides a fast and reliable reCAPTCHA v2 solving service based on image click recognition. Whether you need to bypass reCAPTCHA v2 image grids with cars, traffic lights, crosswalks or any other supported object, the API returns precise click coordinates with high accuracy and low latency. Two response modes, coordinates and cell numbers, give you full flexibility in how you integrate the solution.

The image click solver works with all major browser automation frameworks, including Selenium, Puppeteer and Playwright, and closely mimics real human behavior by clicking the correct tiles instead of injecting tokens. Cap.Guru handles image recognition and object detection behind the scenes, so you can focus on your automation pipeline. With competitive pricing, fast responses and support for all standard reCAPTCHA v2 image challenges, Cap.Guru is a reliable choice for developers who need to solve reCAPTCHA 2 automatically at any scale.