User-Agent reduction

User-Agent reduction limits passively shared browser data to reduce the volume of sensitive information which leads to fingerprinting. The reduction is now complete.

Published on Updated on

Translated to: 日本語

What is User-Agent reduction?

User-Agent (UA) reduction minimizes the identifying information shared in the User-Agent string, which may be used for passive fingerprinting. Now that these changes have been rolled out for general availability, all resource requests have a reduced User-Agent header. As a result, the return values from certain Navigator interfaces are reduced, including: navigator.userAgent, navigator.appVersion, and navigator.platform.

Web developers should review their site code for usage of the User-Agent string. If your site relies on parsing the User-Agent string to read the device model, platform version, or full browser version, you'll need to implement the User-Agent Client Hints API.

Key Term

The User-Agent string is an HTTP request header that allows servers and networks to identify the application, operating system (OS), vendor, and / or version of a user agent. Historically, the User-Agent was shared on every HTTP request and exposed in JavaScript.

User-Agent Client Hints (UA-CH)

User-Agent Client Hints allow access to the full set of User-Agent data, but only when servers actively declare an explicit need for specific pieces of data.

By removing passively exposed user data, we better measure and reduce the amount of information that is intentionally exposed by request headers, JavaScript APIs, and other mechanisms.

Why do we need reduced UA and UA-CH?

Historically, the User-Agent string would broadcast a large string of data about a user's browser, operating system, and version with every HTTP request. This was problematic for two reasons:

  • The granularity and abundance of detail can lead to user identification.
  • The default availability of this information can lead to covert tracking.

Reduced UA and UA-CH improve user privacy by sharing only basic information by default.

The reduced User-Agent includes the browser's brand and a significant version, where the request came from (desktop or mobile), and the platform. To access more data, User-Agent Client Hints allow you to request specific information about the user's device or conditions.

Further, over time the User-Agent string grew longer and more complex, which led to error-prone string parsing. UA-CH provides structured and reliable data that is easier to interpret. Existing code that parses the UA string shouldn't break (though it will return less data), and you'll need to migrate to UA-CH if your site needs specific client information.

How does the reduced UA and UA-CH work?

Here is a brief example of how the reduced User-Agent string and UA-CH work. For a more in-depth example, review Improving user privacy and developer experience with User-Agent Client Hints.

A user opens the browser and enters into the address bar:

  1. The browser sends a request to load the webpage.
    1. The browser includes the User-Agent header with the reduced User-Agent string. For example: User-Agent: Mozilla/5.0 (Linux; Android 10; K) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/ Mobile Safari/537.36
    2. The browser includes that same information in the default User-Agent Client Hint headers. For example:
      Sec-CH-UA: "Chrome"; v="98"
      Sec-CH-UA-Mobile: ?1
      Sec-CH-UA-Platform: "Android"
  2. The server can ask the browser to send additional client hints, such as the device model, with the Accept-CH response header. For example: Accept-CH: Sec-CH-UA, Sec-CH-UA-Mobile, Sec-CH-UA-Platform, Sec-CH-UA-Model
  3. The browser applies policies and user configuration to determine what data is allowed to return to the server in subsequent request headers. For example:
    Sec-CH-UA: "Chrome"; v="93"
    Sec-CH-UA-Mobile: ?1
    Sec-CH-UA-Platform: "Android"
    Sec-CH-UA-Model: "Pixel 2"

Critical Client Hints

If you need a specific set of Client Hints in your initial request, you can use the Critical-CH response header. Critical-CH values must be a subset of the values requested by Accept-CH.

For example, the initial request may include a request for Device-Memory and Viewport-Width, where Device-Memory is considered critical.

GET / HTTP/1.1

HTTP/1.1 200 OK
Content-Type: text/html
Accept-CH: Device-Memory, Viewport-Width
Vary: Device-Memory, Viewport-Width
Critical-CH: Device-Memory

If the browser requires a critical hint (Critical-CH) to properly render the webpage, then the server can ask for this additional information with the Accept-CH header. Then, the browser can send a new request for the page, including the critical hint.

In summary, Accept-CH requests all values you'd like for the page, while Critical-CH requests only the subset of values you must have on-load to properly load the page. Refer to the Client Hints Reliability specification for more information.

Detect tablet devices with the UA-CH API

As the line between mobile, tablet, and desktop devices continues to become less distinct and dynamic form factors are more common (folding screens, switching between laptop and tablet mode), it's advisable to use responsive design and feature detection to present an appropriate user interface.

However, information provided by the browser for both the User-Agent string and User-Agent Client Hints comes from the same source, so the same forms of logic should work.

For example, if this pattern is checked on the UA string:

  • Phone pattern: 'Android' + 'Chrome/[.0-9]* Mobile'
  • Tablet pattern: 'Android' + 'Chrome/[.0-9]* (?!Mobile)'

The matching default UA-CH headers interface may be checked:

  • Phone pattern: Sec-CH-UA-Platform: "Android", Sec-CH-UA-Mobile: ?1
  • Tablet pattern: Sec-CH-UA-Platform: "Android", Sec-CH-UA-Mobile: ?0

Or the equivalent JavaScript interface:

  • Phone pattern: navigator.userAgentData.platform === 'Android' && === true
  • Tablet pattern: navigator.userAgentData.platform === 'Android' && === false

For hardware-specific use cases, the device model name can be requested via the high-entropy Sec-CH-UA-Model hint.

How do I use and test reduced UA?

To begin, review your site code for instances and uses of the User-Agent string. If your site relies on parsing the User-Agent string to read the device model, platform version, or full browser version, you'll need to implement the UA-CH API.

Once you've updated to the UA-CH API, you should test to ensure you get the data you expect from the User-Agent. There are three ways to test, each increasing in complexity.

Scaled availability for User-Agent reduction means the fully reduced UA string shipped on all Chrome devices. Reduction began with a Chrome minor release in Q2 of 2022.

Test the string locally

There are a couple of methods to test the reduced User-Agent locally:

  • Enable the chrome://flags/#reduce-user-agent flag.
    • This will set your local browser to receive just the reduced user-agent string for all sites, before it becomes the default setting.
  • Configure an emulated device in DevTools with the right user-agent string and client hints.
    • In the top right of DevTools, click Settings > Devices > Add custom device... to configure an emulated device with any combination of user-agent string and User-Agent Client Hints values you need.
    • In the top left of DevTools, click Toggle Device Toolbar to open the DevTools UI to emulate a device.
  • Launch Chrome with the --user-agent="Custom string here".
    • Use this command-line flag to run Chrome with a custom User-Agent string. More on command-line flags here.

Transform the string in your site's code

If you process the existing Chrome user-agent string in your client-side or server-side code, you can transform that string to the new format to test compatibility. You can test by either overriding and replacing the string, or generating the new version and test side by side.

Review these User-Agent reduction snippets for example regular expressions.

Support for Client Hints and critical hints

There are three default Client Hints returned to the server, including browser name and major version, a boolean that indicates if the browser is on a mobile device, and the operating system name. These are sent after the Transport Layer Security protocol (TLS) handshake. These are already available and supported in your browser.

However, there may be times when you need to retrieve critical information for your site to render.

Optimize critical hints


Using critical hints should be rare, so make sure you've reviewed the reason for implementation. The question to ask yourself is, do you require extended data on the initial page load? Will your page fail to load without this information?

A TLS handshake is the first step to create a secure connection between the browser and web server. Without an intervention, the Critical-CH response header was designed to tell the browser to immediately retry the request if the first one was sent without a critical hint.

Sequence diagram for Client Hints with critical hints
When a critical hint is requested by the server, the client will retry sending the first request for the webpage with the critical hint. In this example, the hint for Sec-CH-UA-Model is requested twice: once as a Client Hint with Accept-CH and again as a critical hint with Critical-CH.

To optimize critical hints (Critical-CH header), you must intercept this handshake and provide a model for Client Hints. These steps may be complex, and require advanced knowledge.

The ACCEPT_CH HTTP/2 and HTTP/3 frames, combined with the TLS ALPS extension, is a connection-level optimization to deliver the server’s Client Hint preferences in time for the first HTTP request. These require complex configuration, and we recommend only using this for truly critical information.

BoringSSL (a fork of OpenSSL) helps you work with Google’s experimental features in Chromium. At this time, ALPS is only implemented in BoringSSL.

If you need to use critical hints, refer to our guide on critical hints reliability and optimization.


How long will hints specified via the Accept-CH header be sent?

Hints specified via the Accept-CH header will be sent for the duration of the browser session or until a different set of hints is specified.

Does UA-CH work with HTTP/2 and HTTP/3?

UA-CH works with both HTTP/2 and HTTP/3 connections.


Client Hints are only sent over secure connections, so make sure your uses HTTPS.

Do subdomains (and CNAMEs) require a top-level page Permissions-Policy to access high-entropy UA-CH?

High-entropy UA-CH on request headers are restricted on cross-origin requests regardless of how that origin is defined on the DNS side. Delegation must be handled via Permissions-Policy for any cross-origin subresource or obtained via JavaScript that executes in the cross-origin context.

How does User-Agent reduction affect bot detection?

Chrome's change to its User-Agent string does not directly impact the User-Agent string that a bot chooses to send.

Bots may choose to update their own strings to reflect the reduced information Chrome sends, but that is entirely their implementation choice. Chrome is still sending the same User-Agent format, and bots that append their own identifier to the end of a Chrome User-Agent string can continue to do so.

For any concerns with specific bots, it may be worth reaching out directly to the owners to ask if they have any plans to change their User-Agent string.

Engage and share feedback

Find out more

Updated on Improve article

This site uses cookies to deliver and enhance the quality of its services and to analyze traffic. If you agree, cookies are also used to serve advertising and to personalize the content and advertisements that you see. Learn more about our use of cookies.