Hey there! We've found the following might help you get unblocked faster:

• 🧵 Mismatch between browserless chrome and playwright

• 🧵 Update Browserless Playwright to 1.4.4

• 🧵 Running Playwright using Browserless-v2 template on Railway

If you find the answer from one of these, please let us know by solving the thread!

Fixed: Playwright–Browserless Version Mismatch

You're getting that error because the Playwright client you're using (v1.55) doesn't match the Browserless server version (v1.41), and the WebSocket /playwright endpoint requires both to be in sync.

There are two solid options to fix this:

Option 1 – Downgrade Playwright

If you're set on using the /playwright WebSocket endpoint, you have to match the server version. That means downgrading your local Playwright package to v1.41 (since that’s what the server is running based on your logs). Example:

npm install playwright@1.41.0
# or for Python:
pip install playwright==1.41.0

That should clear up the 428 version mismatch error.

BUT if you're open to a simpler fix…

Option 2 – Use CDP Instead (Recommended)

Browserless supports connecting via Chrome DevTools Protocol (CDP), which doesn’t require the Playwright versions to match. This is the cleanest way around the issue.

Just remove /playwright from the URL and connect like this:

JavaScript/Node:

const browser = await chromium.connectOverCDP("wss://your-browserless-url?token=your-token");

Python:

browser = playwright.chromium.connect_over_cdp("wss://your-browserless-url?token=your-token")

The cool part here is that CDP is version-tolerant—no more client/server mismatch headaches.

For Crawl4AI

Since you mentioned you're running Crawl4AI, you can set the connection mode to use CDP directly:

from crawl4ai import BrowserConfig, AsyncWebCrawler
import os

browser_cfg = BrowserConfig(
    browser_type="chromium",
    headless=True,
    browser_mode="cdp",
    cdp_url=os.environ["BROWSER_WS_ENDPOINT"]  # This should be the CDP URL (no `/playwright`)
)
crawler = AsyncWebCrawler(config=browser_cfg)

Just make sure your Railway environment has the correct BROWSER_WS_ENDPOINT set without the /playwright path.

TL;DR:
Either downgrade to Playwright v1.41 (if you’re sticking with the /playwright WebSocket), or better yet, switch to CDP which is way more forgiving on versions. I’d go with CDP if possible—it’s what Browserless recommends too.

Hope that helps—should get you past that error and up and crawling.

Hi @bytekim, this has resolved the issue for my crawl4ai instance.

Thank you so much!