Developer 🔒 Runs in your browser — no data sent to servers

User-Agent Parser

Read the browser, operating system and device in any user-agent string, see your own, and recognise search engine and AI crawlers.

Examples:

How to Use

1. Your own user agent is filled in when the page opens; press Use my user agent to bring it back.
2. Or paste any user-agent string, for example from a server log, or pick one from Examples.
3. Read the browser, engine, operating system and device, and the crawler or tool when the string belongs to one.
4. Read the notes: they explain what a string cannot tell, such as Windows 11 or the real macOS version.
5. In Chromium-based browsers, the Client Hints table shows what this browser reports beyond its user agent.

The string is parsed in your browser and not sent anywhere.

Frequently Asked Questions

What is a user agent?

A line of text a browser or program sends with every request to say what it is: the browser, its version, the engine and the operating system. Servers use it for statistics, bug workarounds and to recognise crawlers.

Why does my Windows 11 PC show Windows 10?

Browsers send Windows NT 10.0 for both Windows 10 and 11. Only Client Hints tell them apart; the tool reads them in browsers that support them.

Why do all browsers start with Mozilla/5.0?

For historical compatibility. Early sites only sent full pages to browsers that said Mozilla, so every browser since has copied the prefix, and later also names such as AppleWebKit and Safari.

How do I know a request really came from Googlebot?

Not from the user agent, which anyone can copy. Look up the IP address with reverse DNS: it must end in googlebot.com or google.com, and a forward lookup of that name must return the same IP.

Which AI crawlers does it recognise?

Among others GPTBot, OAI-SearchBot and ChatGPT-User, ClaudeBot and Claude-User, PerplexityBot, CCBot, Bytespider, Meta-ExternalAgent and Amazonbot, each labelled as training, AI search or fetching a page for a user.

User-Agent Parser - Read Any User-Agent String

Updated September 30, 2026 · By Saman

Every browser, crawler and script sends a user-agent string with each request. This tool turns that string into plain facts: the browser and its version, the rendering engine, the operating system and the kind of device. It shows your own user agent as soon as the page opens, and it recognises search engine crawlers, AI crawlers, link-preview bots and command-line tools.

It also says what a user agent can no longer tell you. Browsers now freeze or trim parts of the string for privacy, so the notes explain when the Windows version, the macOS version or the phone model cannot be read, and the Client Hints table shows what your browser reports instead.

How to use it

Your own user agent is filled in; Use my user agent brings it back after you change the text.
Paste any other string, for example a line from a server log, or choose one from Examples.
Read the browser, engine, operating system and device, and the crawler or tool if there is one.
Read the notes under the results, and the Client Hints table if your browser is based on Chromium.

Anatomy of a user-agent string

A typical Chrome string on Windows reads: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/129.0.0.0 Safari/537.36. Only a few parts carry information:

Part	What it says
`Mozilla/5.0`	Nothing today; every browser sends it for historical compatibility
`Windows NT 10.0; Win64; x64`	Operating system and processor architecture
`AppleWebKit/537.36 (KHTML, like Gecko)`	Engine heritage, also kept for compatibility
`Chrome/129.0.0.0`	The browser and its major version
`Safari/537.36`	Another compatibility token; it does not mean Safari

Because of this layering, the order of checks matters: Edge also says Chrome and Safari, and Chrome also says Safari. The parser checks the most specific names first.

What the string no longer tells you

Windows 11: it sends Windows NT 10.0, the same as Windows 10.
macOS: Safari, Chrome and Firefox freeze the version at 10_15_7.
Android: Chrome sends Android 10; K whatever the real version and phone model.
Browser version: Chromium-based browsers send only the major version, followed by .0.0.0.
iPad: in desktop mode it sends the same string as a Mac.

User-Agent Client Hints replace these details. A site can ask for them, and the browser answers with the platform version, the model and the full browser version; that is how the table on this page can tell Windows 11 from Windows 10. Safari and Firefox do not support Client Hints.

Crawlers and AI bots

The parser recognises the main search engine crawlers (Googlebot, Bingbot, YandexBot, Baiduspider, DuckDuckBot, Applebot), AI crawlers, link-preview bots from social networks and messaging apps, SEO crawlers and tools such as curl, Wget and Python Requests. AI crawlers fall into three groups, and the tool labels each one:

Group	Examples	What it does
Training	GPTBot, ClaudeBot, CCBot, Bytespider, Meta-ExternalAgent	Collects pages to train AI models
AI search	OAI-SearchBot, Claude-SearchBot, PerplexityBot	Indexes pages so an AI search engine can cite them
User-triggered	ChatGPT-User, Claude-User, Perplexity-User	Fetches one page because a person asked an assistant about it

Google-Extended and Applebot-Extended are different: they are only names for robots.txt rules and never appear in a request. Googlebot and Applebot do the crawling, and those rules decide whether the pages may also be used for AI training.

A user agent can be faked by anyone. Before you trust or block a crawler by name, check its IP address: a reverse DNS lookup for search engines, or the IP ranges the operator publishes. Rules in robots.txt are addressed by these names too, for example User-agent: GPTBot.

Related tools

To check a crawler's network, look up its range in the subnet calculator. For a user agent that arrives URL-encoded in a log, decode it first with the URL encoder. The task page what is my user agent shows only your own browser.