Googlebot user agent strings: smartphone, desktop and Chrome
The exact Googlebot user agent strings for smartphone and desktop, what Chrome/W.X.Y.Z means, and how to match Googlebot in robots.txt, logs and WAF rules.
By Crawlito · 7 min read
In short
- Googlebot sends two main user agent strings, Smartphone and Desktop, and both contain the substring 'Googlebot/2.1; +http://www.google.com/bot.html'.
- Chrome/W.X.Y.Z in Google's documented strings is a placeholder for the Chromium version Googlebot uses, which increases over time, so Google says to use wildcards for the version number.
- Both Googlebot types obey the same robots.txt token, Googlebot, so robots.txt cannot target Googlebot Smartphone or Googlebot Desktop separately.
- For most sites, Google says the majority of Googlebot requests come from the smartphone crawler, because Google Search primarily indexes the mobile version of content.
- A user agent match is only a claim: confirm Googlebot with a reverse and forward DNS lookup to googlebot.com, or by checking the IP against Google's common-crawlers.json list.
On this page
- The current Googlebot user agent strings
- What Chrome/W.X.Y.Z means and why exact matches break
- How to tell Googlebot Smartphone from Desktop in logs
- Which token to use in robots.txt and which in log searches
- How to confirm a Googlebot match is real
- A Googlebot firewall rule that survives Chrome updates
- Frequently asked questions
- Check what Googlebot gets from your pages
Googlebot uses two main user agent strings: a smartphone string that starts with an Android 6.0.1 Nexus 5X browser, and a desktop string with no operating system at all. Both contain Googlebot/2.1; +http://www.google.com/bot.html and a Chrome/W.X.Y.Z version that Google changes as it updates its rendering engine. So match on the token Googlebot, never on a full string, and confirm the IP before you trust it.
This post gives the exact strings from Google's documentation, explains the Chrome placeholder, and shows which pattern to use in robots.txt, log searches and firewall rules. For the other crawlers you may want to recognise, see the AI crawler user agents list.
The current Googlebot user agent strings
Google lists these in its common crawlers reference, last updated 30 October 2026 at the time of writing.
Googlebot Smartphone
Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/W.X.Y.Z Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)
Googlebot Desktop
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; Googlebot/2.1; +http://www.google.com/bot.html) Chrome/W.X.Y.Z Safari/537.36Google also lists two strings it says are used rarely:
Mozilla/5.0 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)
Googlebot/2.1 (+http://www.google.com/bot.html)The differences that matter for filtering:
| Variant | Distinctive parts | Contains Googlebot/2.1 |
|---|---|---|
| Smartphone | Android 6.0.1; Nexus 5X, Mobile Safari | Yes, at the end |
| Desktop | No OS token, Googlebot inside the (KHTML, like Gecko; ...) brackets | Yes, in the middle |
| Rare, compatible | Mozilla/5.0 (compatible; ...), no Chrome version | Yes |
| Rare, bare | Starts with Googlebot/2.1 | Yes |
The position of the token differs between the smartphone and desktop strings, so any pattern that assumes Googlebot comes last, or that the string starts with Mozilla/5.0 (Linux, will miss one of them.
What Chrome/W.X.Y.Z means and why exact matches break
W.X.Y.Z is not literal text. Google's note on the same page says it is a placeholder for the version of Chrome the user agent uses, gives 41.0.2272.96 as an example, and says the number "increases over time to match the latest Chromium release version used by Googlebot". Google's advice is direct: when you search logs or filter a server, use wildcards for the version number rather than an exact version.
That is why rules copied from an old forum post stop working. A rule like this matches one Googlebot release and nothing after it:
# Brittle: breaks the next time Googlebot's Chromium version changes
if ($http_user_agent = "Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.96 Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)") {
set $is_googlebot 1;
}If that rule is part of an allowlist, the day the version changes is the day real Googlebot falls through to your default action, which may be a challenge or a block. The Googlebot 403 guide covers how that shows up in Search Console.
A pattern that survives updates matches the token and ignores the version:
map $http_user_agent $claims_googlebot {
default 0;
"~*googlebot" 1;
}If you really need to pin the shape of the string, wildcard the version instead of deleting it: Chrome/[0-9.]+ in a regex. Either way, $claims_googlebot means only that the request says it is Googlebot. The IP check comes later in this post.
How to tell Googlebot Smartphone from Desktop in logs
Google's Googlebot page says you can identify the subtype from the user agent header, and that for most sites the majority of Googlebot requests use the mobile crawler, because Google Search primarily indexes the mobile version of content. A large desktop share is not wrong in itself, but it is worth a look if you serve different content to each, as described in mobile and desktop content parity.
This splits Googlebot lines in a standard combined log by subtype:
grep -i 'googlebot/2\.1' /var/log/nginx/access.log | awk '
/Android/ && /Mobile/ { mobile++; next }
/Chrome\// { desktop++; next }
{ other++ }
END { printf "smartphone %d\ndesktop %d\nother %d\n", mobile, desktop, other }'Matching googlebot/2.1 rather than googlebot keeps Googlebot-Image and Googlebot-Video out of the count. The "other" line catches the two rare strings that carry no Chrome version. As with any log grep, a line also matches if the text appears in the URL or referrer, and every count is a count of claims until verified.
Which token to use in robots.txt and which in log searches
robots.txt and log filters need different patterns, and mixing them up is a common source of mistakes.
In robots.txt: Googlebot, nothing more
Google documents one robots.txt token for both subtypes: Googlebot. Its Googlebot page states that both crawler types obey the same token, so you cannot selectively target Googlebot Smartphone or Googlebot Desktop with robots.txt.
User-agent: Googlebot
Disallow: /internal-search/Writing Googlebot/2.1 adds nothing: Google's robots.txt specification says non-matching text is ignored, so googlebot/1.2 and googlebot* are both equivalent to googlebot. Do not paste the full user agent string into a User-agent: line, though. It starts with Mozilla/5.0, not with the token, and RFC 9309 defines the value as a product token made only of letters, underscores and hyphens. The RFC also requires case-insensitive matching of that token, so googlebot and Googlebot are the same group.
Two consequences of Google's group rules are worth knowing. A crawler follows only the most specific group that matches it, and specific groups are not combined with *. And several Google crawlers list Googlebot as a second token, among them Googlebot-Image, Googlebot-Video, Googlebot-News and Google-InspectionTool. Google says a crawler needs to match only one of its tokens for a rule to apply, so when no group names them directly, a Googlebot group applies to them too. The token details are in Googlebot-Image, Video and News tokens.
In logs and firewalls: know what else contains "Google"
A case-insensitive googlebot substring catches Googlebot Smartphone, Desktop, both rare strings, Googlebot-Image/1.0 and Googlebot-Video/1.0. It does not catch Google's other agents, several of which reuse the same Nexus 5X browser prefix:
| Agent | Distinctive token in the string | Matches googlebot? |
|---|---|---|
| Googlebot (both subtypes) | Googlebot/2.1 | Yes |
| Googlebot-Image, Googlebot-Video | Googlebot-Image/1.0, Googlebot-Video/1.0 | Yes |
| Google-InspectionTool | Google-InspectionTool/1.0 | No |
| Storebot-Google | Storebot-Google/1.0 | No |
| GoogleOther | GoogleOther | No |
| AdsBot-Google-Mobile | AdsBot-Google-Mobile | No |
| Google-Agent | Google-Agent | No |
The strings come from Google's pages for common crawlers, special-case crawlers and user-triggered fetchers. Googlebot-News has no string of its own: Google says it crawls with the regular Googlebot strings. If a firewall rule allows only googlebot, URL Inspection tests in Search Console arrive as Google-InspectionTool and are not covered, which is explained in the Google-InspectionTool guide.
How to confirm a Googlebot match is real
Google's own Googlebot page warns that the user agent header is often spoofed by other crawlers. Its verification guide gives two methods for common crawlers such as Googlebot:
- Reverse and forward DNS. The source IP should reverse-resolve to a hostname matching
crawl-***-***-***-***.googlebot.comorgeo-crawl-***-***-***-***.geo.googlebot.com, and a forward lookup of that hostname must return the same IP. - IP ranges. Match the IP against
common-crawlers.json, which Google publishes in CIDR format.
host 35.247.243.240
# 240.243.247.35.in-addr.arpa domain name pointer geo-crawl-35-247-243-240.geo.googlebot.com.
host geo-crawl-35-247-243-240.geo.googlebot.com
# geo-crawl-35-247-243-240.geo.googlebot.com has address 35.247.243.240That is Google's second example, and it shows the geo.googlebot.com form, which a check written only for crawl-*.googlebot.com would reject. The full scripts, including a forward-confirmed DNS function and a JSON range checker, are in how to verify real crawlers, so they are not repeated here.
One location detail affects geo rules: Google's crawler overview says Google egresses primarily from US IP addresses, and may try IPs in other countries if it detects a site blocking requests from the US. A country allowlist that excludes the US can therefore push Googlebot onto addresses you did not expect.
A Googlebot firewall rule that survives Chrome updates
Put the pieces together and a durable rule has three parts:
- Match the claim loosely. Use a case-insensitive
googlebotsubstring, never a full string or a pinned Chrome version. - Decide on the IP, not the string. Allow the request only if the source IP is in
common-crawlers.jsonor passes the DNS check. Treat agooglebotclaim from any other IP as an impostor. - Refresh the IP list on a schedule. Download the JSON automatically instead of pasting ranges, and keep the previous file if a download fails.
Behind a CDN, log and check the forwarded client IP rather than the proxy's address, or every request will fail step 2. If your edge provider already labels verified bots, as Cloudflare does, use that signal for step 2 instead of maintaining your own list.
This matters beyond classic search. Google says robots.txt rules for Googlebot are the control for how sites are crawled for Search, and that AI is built into Search, in its page on AI features and your website. A rule that drops real Googlebot after a Chromium update affects every Search feature at once, AI Overviews and AI Mode included.
Frequently asked questions
What is the Googlebot user agent string?
Googlebot Smartphone sends Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/W.X.Y.Z Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html). The desktop string drops the Android part and moves the Googlebot token inside the KHTML brackets. W.X.Y.Z stands for the current Chrome version.
Can I block Googlebot Desktop but allow Googlebot Smartphone in robots.txt?
No. Google says both subtypes obey the same Googlebot token, so robots.txt cannot target one of them. Telling them apart from the user agent is possible only at the server, and serving them different content brings its own risks.
What Chrome version does Googlebot use right now?
Google does not publish a fixed number in its reference: it shows the placeholder W.X.Y.Z and says the version increases over time to match the latest Chromium release Googlebot uses. Read it from your own recent log lines if you need it, and never hard-code it in a rule.
Is every request with Googlebot in the user agent from Google?
No. Anyone can send the string, and Google warns that it is often spoofed. Trust the claim only when the IP passes the reverse and forward DNS check or appears in common-crawlers.json, which covers Google's common crawlers, Googlebot included.
Check what Googlebot gets from your pages
If you changed a filter or firewall rule that mentions Googlebot, test the result rather than the rule. As a next step, run the free check on one of your pages: it checks a single page and needs no account.
Filed under AI crawlers. Topics: googlebot, google, server-logs, waf, ai-crawlers