User Agent Parser

How to use the User Agent Parser

1. Look at Your browser to see what your own browser reports, including client hints such as the real Windows or macOS version where supported.
2. Paste a user agent string from a log, an analytics report or a bug ticket into the parser, or click a sample to try it.
3. Read the browser and version, rendering engine, operating system, device type and device model.
4. Paste several user agents, one per line, to get a quick table that identifies each one.

Features

• Shows your own user agent and high-entropy client hints
• Detects Chrome, Safari, Firefox, Edge, Samsung Internet, Opera and in-app browsers
• Identifies search engine, AI, SEO and social media crawlers
• Recognizes command-line tools and HTTP libraries such as curl and Python Requests
• Phone, tablet, desktop, TV and console detection with device models
• Parses a whole list of user agents at once

What a user agent string is

Every time a browser requests a page, it sends a User-Agent header describing itself. Servers, analytics tools and logs use it to tell which browser, operating system and device made the request. The string is free text with loose conventions, which is why it needs a parser to read reliably.

Anatomy of a user agent

Here is a typical Chrome on Windows string, split into its parts:

Why does Chrome claim to be Mozilla, WebKit, Gecko and Safari at once? In the early web, sites served their best pages only to browsers they recognized, so each new browser copied the tokens of the popular ones to avoid being shut out. The habit stuck. As a result, the order of the checks matters: Edge, Opera and Samsung Internet all contain Chrome/, so a parser must look for their own tokens (Edg/, OPR/, SamsungBrowser/) first.

Frozen and reduced user agents

User agent strings reveal enough detail to help fingerprint and track people, so browsers now hide some of it. Chrome reports a fixed minor version (130.0.0.0), shows Android 10; K instead of the real Android version and phone model, and reports Windows NT 10.0 on Windows 11. On a Mac, Safari and Chrome report macOS 10_15_7 and Firefox reports 10.15, whatever the actual version. iPads ask for desktop sites by default and send a Mac user agent. A parser can only report what the string contains, so treat those values as approximate.

Client hints: the modern replacement

Chromium-based browsers (Chrome, Edge, Opera, Samsung Internet) support User-Agent Client Hints. A page can read basic values, such as the brand, whether the device is mobile and the platform, from navigator.userAgentData or the Sec-CH-UA request headers. Detailed values, such as the full browser version, platform version, CPU architecture and device model, must be requested explicitly. That is how the panel above can tell Windows 11 from Windows 10. Firefox and Safari do not support client hints.

Bots, crawlers and tools

Search engines, AI companies, SEO services and social networks run crawlers that identify themselves in the user agent, such as Googlebot, bingbot, GPTBot or facebookexternalhit. Command-line tools and libraries send simpler strings like curl/8.7.1 or python-requests/2.32.3. Remember that anyone can send any user agent. Google and Bing recommend confirming their crawlers with a reverse DNS lookup on the requesting IP address; for robots.txt rules aimed at specific crawlers, see the robots.txt generator.

Use feature detection, not sniffing

User agents are great for analytics, debugging and spotting bots, but poor for deciding what code to run. Testing for a feature directly, for example with 'share' in navigator or CSS @supports, keeps working when browsers change their strings and does not break for browsers you did not anticipate.

Frequently asked questions