Impersonate guide

Supported browser versions

Supported impersonate versions, as supported by our fork of curl-impersonate:

However, only Chrome-like browsers are supported. Firefox support is tracked in #59.

  • chrome99

  • chrome100

  • chrome101

  • chrome104

  • chrome107

  • chrome110

  • chrome116 1

  • chrome119 1

  • chrome120 1

  • chrome99_android

  • edge99

  • edge101

  • safari15_3 2

  • safari15_5 2

  • safari17_0 1

  • safari17_2_ios 1


  1. Added in version 0.6.0.

  2. Fixed in version 0.6.0, previous http2 fingerprints were not correct.

Which version to use?

Generally speaking, you should use the latest Chrome or Safari versions. As of 0.6, they’re chrome120, safari17_0 and safari17_2_ios. To always impersonate the latest avaiable browser versions, you can simply use chrome, safari and safari_ios.

from curl_cffi import requests

requests.get(url, impersonate="chrome")

iOS has restrictions on WebView and TLS libs, so safari_x_ios should work for most apps. If you encountered an android app with custom fingerprints, you can try the safari ios fingerprints given that this app should have an iOS version.

How to customize my fingerprints? e.g. okhttp

It’s not fully implemented, yet.

There are many parts in the JA3 and Akamai http2 fingerprints. Some of them can be changed, while some can not be changed at the moment. The progress is tracked in

To modify them, use curl.setopt(CurlOpt, value), for example:

from curl_cffi import Curl, CurlOpt, requests

c = Curl()
c.setopt(CurlOpt.HTTP2_PSEUDO_HEADERS_ORDER, "masp")

# or
requests.get(url, curl_options={CurlOpt.HTTP2_PSEUDO_HEADERS_ORDER, "masp"})

Here are a list of options:

For TLS/JA3 fingerprints:

and non-standard TLS options created for this project:






For Akamai http2 fingerprints, you can fully customize the 3 parts:

  • CURLOPT_HTTP2_PSEUDO_HEADERS_ORDER, sets http2 pseudo header order, for example: masp (non-standard HTTP/2 options created for this project).

  • CURLOPT_HTTP2_SETTINGS sets the settings frame values, for example 1:65536;3:1000;4:6291456;6:262144 (non-standard HTTP/2 options created for this project).

  • CURLOPT_HTTP2_WINDOW_UPDATE sets intial window update value for http2, for example 15663105 (non-standard HTTP/2 options created for this project).

Should I randomize my fingerprints for each request?

You can use a random from the list above, like:

random.choice(["chrome119", "chrome120", ...])

However, be aware of the browser market share, very old versions are not good choices.

Generally, you should not try to generate a customized random fingerprints. The reason is that, for a given browser version, the fingerprints are fixed. If you create a new random fingerprints, the server is easy to know that you are not using a typical browser.

If you were thinking about ja3, and not ja3n, then the fingerprints is already randomized, due to the extension permutation feature introduced in Chrome 110.

AFAIK, most websites use an allowlist, not a blocklist to filter out bot traffic. So I don’t think random ja3 fingerprints would work in the wild.

Can I change JS fingerprints with this library?

No, you can not. As the name suggests, JavaScript fingerprints are generated using Javascript APIs provided by real browsers. curl_cffi is a python binding to a C library, with no browser or JavaScript runtime under the hood.

If you need to impersonate browsers on the JavaScript perspective, you can search for “Anti-detect Browser”, “Playwright stealth” and similar keywords. Or simply use a commercial plan from our sponsors.