# ClaudeBot vs GPTBot: two training crawlers, one you can slow down

> For the complete documentation index, see [llms.txt](https://attensira.com/llms.txt).
> Every page on this site is also served as Markdown: append `.md` to any URL, or send `Accept: text/markdown`.

> ClaudeBot and GPTBot both collect training data. How Anthropic and OpenAI differ on crawl-delay, verification and what blocking does, with our own logs.

- Canonical page: https://attensira.com/compare/claudebot-vs-gptbot
- Family: compare
- Last updated: 2026-10-04
- Author: Karl-Gustav Kallasmaa (https://attensira.com/about)
- Claims as data: https://attensira.com/claims.json?slug=claudebot-vs-gptbot

**In short:** ClaudeBot and GPTBot do the same job for different labs: they collect web content that may be used to train Anthropic's and OpenAI's models. Blocking either keeps your future pages out of that lab's training data and, per both vendors, does not remove you from their answers, which use separate search and user agents. Decide both together.

**Readings, 3 Sep 2026 to 4 Oct 2026**

| Measure | Value |
| --- | --- |
| our most-fetched pages read by both GPTBot and ClaudeBot (n=30 pages) | 16 of 30 |
| read by GPTBot but not ClaudeBot; none the other way round (n=30 pages) | 14 of 30 |
| gap in root-block share, GPTBot over ClaudeBot (n=8,036 domains) | 0.9-point |

_attensira.com AI traffic log, top 30 pages, 5 Sep-4 Oct 2026, read over the Attensira MCP 2026-10-04, by claimed user agent; robots.txt of the Tranco top-10,000 (list K9QPW), read 2026-09-03 · 4 Oct 2026_

## Verdict

Same job, two labs, slightly different controls. Both are training crawlers that obey robots.txt and have no documented effect on live answers. Anthropic adds Crawl-delay support; OpenAI adds a per-crawler address list. Most sites should make one training decision and write it for both.

## The readings on this page

### The readings at the top

attensira.com AI traffic log, top 30 pages, 5 Sep-4 Oct 2026, read over the Attensira MCP 2026-10-04, by claimed user agent; robots.txt of the Tranco top-10,000 (list K9QPW), read 2026-09-03 · 3 Sep 2026 to 4 Oct 2026

| Reading | Value | Sample |
| --- | --- | --- |
| our most-fetched pages read by both GPTBot and ClaudeBot | 16 of 30 | n=30 pages |
| read by GPTBot but not ClaudeBot; none the other way round | 14 of 30 | n=30 pages |
| gap in root-block share, GPTBot over ClaudeBot | 0.9-point | n=8,036 domains |

### What does ChatGPT say about verifying a crawler?

Asked "how to see which AI bots crawl my website" on 4 October 2026, ChatGPT told readers to verify the source IP and named OpenAI's published ranges. It did not mention that Anthropic publishes one too, at claude.com/crawling/bots.json. Quoted as it came.

Stored ChatGPT answer to “how to see which AI bots crawl my website”, asked 4 Oct 2026 (n=1 answer, attensira.com workspace, Attensira MCP).

> Don't rely solely on the user-agent. A bot can pretend to be another bot, so for security/analytics purposes you should also verify the source IP. Google, for example, documents methods for verifying that requests actually originate from Google infrastructure. Likewise, OpenAI publishes IP ranges for its crawlers.

Named, in order: Google, OpenAI. Attensira was not named.

### What does Perplexity tell someone who wants to see these crawlers?

Asked "how to see which AI bots crawl my website" on 4 October 2026, Perplexity put GPTBot and ClaudeBot side by side as tokens to find in your logs and pointed to published IP lists. Logs, not reputation, are where this decision gets checked. Quoted whole.

Stored Perplexity answer to “how to see which AI bots crawl my website”, asked 4 Oct 2026 (n=1 answer, attensira.com workspace, Attensira MCP).

> To see which AI bots are crawling your site, check your server access logs for known bot user-agent strings (e.g., GPTBot, PerplexityBot, ClaudeBot, Googlebot), and correlate those requests with robots.txt rules and published bot IP lists. Use log parsing to identify which bots hit which pages, how often, and with what response codes. If you want automated tagging, enable your CDN/WAF to log bot identifiers and compare against operator lists.

Named, in order: no brand. Attensira was not named.

### Which of our pages did each crawler fetch?

Of the eight attensira.com pages AI crawlers fetched most in the 30 days to 4 October 2026, GPTBot fetched all eight and ClaudeBot seven; only /about was GPTBot's alone. The two training crawls cover nearly the same pages.

attensira.com AI traffic log, page rows of the 30 most-fetched pages, Attensira MCP, attensira.com workspace · 5 Sep 2026 to 4 Oct 2026 · labelled by the user agent each request claimed · n=8 of the 30 most-fetched pages

| Our page | ClaudeBot | GPTBot |
| --- | --- | --- |
| /contact-sales | Fetched | Fetched |
| / | Fetched | Fetched |
| /pricing | Fetched | Fetched |
| /about | Not seen | Fetched |
| /features/prompt-tracking | Fetched | Fetched |
| /features/content-optimization | Fetched | Fetched |
| /features/crawler-logs | Fetched | Fetched |
| /mcp | Fetched | Fetched |

## How do ClaudeBot and GPTBot compare?

Each cell cites a claim in the Sources section below.

| Criterion | ClaudeBot | GPTBot |
| --- | --- | --- |
| **What it is** |  |  |
| Documented purpose | Collects content that could contribute to model training | Crawls content that may be used to train foundation models |
| Published user-agent string | Not published; match the ClaudeBot token | ...compatible; GPTBot/1.4; +https://openai.com/gptbot |
| Published address list | claude.com/crawling/bots.json, one file for all Anthropic bots | openai.com/gptbot.json, one file per crawler |
| Sibling agents for search and user fetches | Claude-SearchBot and Claude-User | OAI-SearchBot and ChatGPT-User |
| **robots.txt control** |  |  |
| robots.txt user-agent token | ClaudeBot | GPTBot |
| Respects robots.txt | Yes · Yes, per Anthropic | Yes · Yes, controlled by its own token |
| Documented Crawl-delay support | Yes · Yes, documented as supported | Not stated · Not documented on the crawler page |
| Documented effect of blocking | Future materials excluded from training datasets | Content should not be used in training |
| **What we measured** |  |  |
| Top-10k domains allowing it by name (study of 3 Sep 2026, n=8,036) | 276 (3.4%) | 349 (4.3%) |
| Top-10k domains that never name it (n=8,036) | 6,683 (83.2%) | 6,540 (81.4%) |

## What is ClaudeBot?

Anthropic's training crawler. It collects web content that could potentially contribute to training Anthropic's generative models, and Anthropic documents robots.txt and Crawl-delay support for it.

## What is GPTBot?

OpenAI's training crawler. It collects content that may be used to train OpenAI's generative foundation models, is controlled by its own robots.txt token and is covered by its own published address file.

## When should you choose ClaudeBot?

Treat ClaudeBot as the training crawler you can slow instead of block. Anthropic supports the non-standard Crawl-delay extension, so a site whose real problem is load can keep content available for training while cutting the request rate.

## When should you choose GPTBot?

Treat GPTBot as the training crawler you can verify most precisely. OpenAI publishes a separate address list for GPTBot alone, so its training traffic can be told apart from OpenAI's search and user agents by address as well as by name.

## Where does Attensira fit?

Whatever it takes to win the answer. On Growth and Business, Attensira shows which crawlers fetch your pages at your edge next to what ChatGPT, Perplexity, Claude, Gemini and Google AI answer and cite.

## What sets ClaudeBot and GPTBot apart?

- Only Anthropic documents Crawl-delay, so ClaudeBot has a middle option between allow and block: slow it and keep it.
- Blocking Anthropic's addresses at the firewall can stop ClaudeBot reading robots.txt at all, so Anthropic's own advice is to decide in robots.txt and use the address list only to verify.
- A file that blocks only the older anthropic-ai token may not express the ClaudeBot decision its author thought it did; Anthropic's current article does not list that token.

## Are ClaudeBot and GPTBot the same kind of crawler?

Yes. Both are training crawlers, one for Anthropic and one for OpenAI. Anthropic says ClaudeBot helps enhance the utility and safety of its generative AI models by collecting web content that could potentially contribute to their training, and that restricting it signals your future materials should be excluded from its training datasets. OpenAI says GPTBot is used to crawl content that may be used in training its generative AI foundation models.

That makes this the simplest comparison on the bot side of our site. There is no difference of purpose to explain. The two crawlers do the same job for different companies, both obey robots.txt, and neither is documented as affecting what the company's assistant says today. The differences are in the controls each lab gives you and in how much each one actually crawls.

## What does blocking each one do?

The same thing, by each vendor's account: it keeps your future pages out of that lab's training data. Neither document describes removing content collected before the block, and neither ties the training crawler to live answers. Each lab runs separate agents for search and for user-triggered fetches: OAI-SearchBot and ChatGPT-User for OpenAI, Claude-SearchBot and Claude-User for Anthropic. Those are the agents that decide whether ChatGPT or Claude can read and cite your pages right now.

So the decision on this page is a values decision, not a visibility one. Do you want your writing in the training data of commercial models? If not, block both. If you do not mind, or you would rather future models know your product well, allow both. A split, one lab allowed and one blocked, is a legitimate choice only if you have a reason that applies to one company and not the other.

## Where do the two labs' controls differ?

In three places.

Crawl-delay. Anthropic supports the non-standard Crawl-delay extension to robots.txt and shows it with ClaudeBot as the example. A site whose real problem is server load can slow ClaudeBot down without excluding anything. OpenAI's crawler page does not document Crawl-delay for GPTBot, so for OpenAI the documented options are allow, disallow, or a disallow scoped to paths.

Address lists. OpenAI publishes a separate list for each crawler, so GPTBot's addresses are in gptbot.json and nowhere else. Anthropic publishes one list for its bots at claude.com/crawling/bots.json. With OpenAI you can tell training traffic from search traffic by address alone. With Anthropic you need the user-agent token as well.

Documentation shape. OpenAI publishes the full user-agent string for each crawler, including a version number. Anthropic's article names its bots by token and gives the address list, but no full strings. Match on the token in both cases; exact-string filters break when a version changes.

## What should your robots.txt say?

For a site that wants to stay out of both labs' training data while staying readable to both assistants:

```
User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /
```

Leave the search and user agents (OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User) out of that group. If load is the concern rather than reuse, a ClaudeBot group with a Crawl-delay line is the softer option:

```
User-agent: ClaudeBot
Crawl-delay: 10
```

Anthropic adds one practical warning: the rule has to be on every subdomain you want covered, and blocking its addresses at the firewall instead can stop the bot reading your robots.txt at all.

## How often do large sites block each one?

On 3 September 2026 we read the robots.txt files of the top 10,000 domains in Tranco list K9QPW and parsed 8,036 of them. 1,147 (14.3%) disallow GPTBot at the root. 1,077 (13.4%) disallow ClaudeBot. The rates are close enough to suggest that most large sites make one training decision and write it for both labs, which is what this page recommends.

Both rates are far above the rates for the same labs' search agents: 5.8% for OAI-SearchBot and 5.5% for Claude-SearchBot in the same read. Large sites have, on the whole, already separated training from search. The full table is in [our study of AI crawler access](/data/ai-crawler-robots-access).

## How much does each one actually crawl?

Very different amounts, at least on our site. Attensira counts crawler requests at the edge on attensira.com, as it does for customers on Growth and Business. In the 30 days to 4 October 2026 the log held 143,601 AI-crawler requests. 7,536 (5.2%) carried OpenAI user agents, across 428 pages, and about 11% of those were classed as training, which is GPTBot. Only 104 (0.07%) carried Anthropic user agents, across 48 pages, and about 65% of those were autonomous crawls, which is ClaudeBot's pattern.

Even allowing for rounding, GPTBot made more than ten times as many requests to our site as ClaudeBot did, and OpenAI's agents as a whole about seventy times as many as Anthropic's. None of these labels is verified: each is the user agent the request claimed, and none was checked against an address list. And they come from one small site. They are a reminder that crawler load is not evenly spread across labs, and that your own logs, not a vendor's reputation, should decide whether load is a real problem.

## What about the older anthropic-ai token?

It still shows up. In the same read of 8,036 large domains, 690 (8.6%) disallow a token called anthropic-ai at the root. Anthropic's current article documents three bots, ClaudeBot, Claude-User and Claude-SearchBot, and does not list anthropic-ai among them. A file that blocks only anthropic-ai may not express the decision its author thought it did. If you find it in yours, add an explicit ClaudeBot group that says what you mean.

The same housekeeping applies on the OpenAI side. Old guides sometimes list ChatGPT-User as a training crawler or treat GPTBot as the agent behind ChatGPT search. Neither is right by OpenAI's current documentation. A robots.txt file is only as current as the last time someone read the vendors' pages.

## Does blocking training hurt how models describe you?

Nobody can show that it does or does not for a single site, and neither lab claims an effect either way. The argument for allowing training is that a model trained on your pages may describe your product more accurately when it answers without searching. The argument for blocking is that your writing is your asset and you have not agreed to its reuse. Both are reasonable. What is not reasonable is presenting either as a measured visibility effect, because no measurement can isolate it.

## Where does this decision usually go wrong?

In the copy-paste. A team finds a list of AI user agents, pastes all of them into one disallow group, and blocks the search and user agents along with the training crawlers. The training decision was deliberate. The search decision was an accident, and it costs the visibility the team wanted.

The second failure is asymmetry by omission: a file written years ago that blocks GPTBot because it was the one name everyone knew, and has never heard of ClaudeBot. Nobody decided to let Anthropic train on the site while keeping OpenAI out. The file simply predates the question.

## What does neither token do?

Neither token removes content already collected. Neither changes how a model already trained describes you. Neither governs the agents that read pages for search or for a person in a conversation. And neither stops a crawler that ignores robots.txt, though both labs say theirs follow it, and Anthropic adds that its bots do not try to bypass CAPTCHAs. If you need certainty that content stays out of every model, the content belongs behind authentication, not behind a robots.txt line.

## Is there a reason to treat the two labs differently?

Occasionally. A company with a commercial relationship with one lab, or a public position on one lab's licensing terms, may reasonably write different rules. A company that wants to be described well by one assistant in particular may decide to allow its training crawler. Load is the other honest reason, and our own log shows how uneven it can be: a site that struggles under OpenAI's request volume may see almost nothing from Anthropic. Those are real reasons. What matters is that the difference is written down, with a date and a name, so it is a decision rather than an accident of when the file was last edited.

## How should you review the decision?

Twice a year, read both vendors' crawler pages again and compare them with the groups in your file. Labs add agents; a policy that was complete when written can miss a token a year later. Pull a week of edge logs and check that requests claiming to be GPTBot or ClaudeBot come from the published addresses, and that the ones you blocked have stopped fetching anything but robots.txt.

Then look at what the decision does not touch: how assistants answer the questions your buyers ask. Neither ChatGPT nor Perplexity named or cited Attensira in any of 60 answers to "how to see which AI bots crawl my website" over the same 30 days, even though both crawlers read our pages. Visibility is decided by search and user agents, by the pages each assistant trusts, and by what those pages say about you.

## FAQ

### Should I block both ClaudeBot and GPTBot?

If your objection is to your content training commercial models, yes: the reasons apply equally to both labs, and blocking one while allowing the other is rarely a deliberate choice. If you have no objection, allow both. The worst outcome is an accidental split, made by copying half of someone else's list.

### Will blocking them hurt my visibility in ChatGPT or Claude?

Not by either vendor's documentation. OpenAI says a site can allow OAI-SearchBot while disallowing GPTBot. Anthropic says blocking ClaudeBot signals that future materials should be excluded from training, and documents separate agents for search (Claude-SearchBot) and for user requests (Claude-User). Live answers depend on those other agents.

### Can I slow them down instead of blocking?

For ClaudeBot, yes: Anthropic supports the non-standard Crawl-delay extension to robots.txt and shows it with ClaudeBot as the example. OpenAI's crawler page does not document Crawl-delay for GPTBot, so for OpenAI the documented controls are allow, disallow by path, or rate limiting at your server.

### How do I verify each one?

GPTBot's addresses are published at openai.com/gptbot.json, a file for that crawler alone. Anthropic publishes one list for its crawlers at claude.com/crawling/bots.json. Check the source address against the list; the user-agent string alone proves nothing.

### Which lab's agents visit more?

On attensira.com, OpenAI's by far: 7,536 requests with OpenAI user agents across 428 pages in the 30 days to 4 October 2026, against 104 with Anthropic user agents across 48 pages. Of OpenAI's, about 11% were classed as training; most were its search crawler. Your own logs are the only answer that applies to you.

### Does blocking remove what they already collected?

No. Both vendors describe the effect of blocking as forward-looking: future materials excluded, content not used in future training. Nothing in either document describes removing pages already collected.

## Sources

Each entry below is one claim on this page. `retrieved` is the date the exporter or a human last saw the source say this, and nothing more: it is not a publication date and not a promise of re-verification. `cite` is the anchor on the canonical page where this claim is rendered.

### claudebot-purpose

Anthropic documents ClaudeBot as collecting web content that could potentially contribute to training its generative models.

- source: https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (support.claude.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-claudebot-purpose
- side: claudebot

> helps enhance the utility and safety of our generative AI models by collecting web content that could potentially contribute to their training
### claudebot-token

ClaudeBot is controlled with the robots.txt user-agent token ClaudeBot.

- source: https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (support.claude.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-claudebot-token
- side: claudebot
### claudebot-robots

Anthropic states that its bots respect do-not-crawl signals by honouring industry standard directives in robots.txt.

- source: https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (support.claude.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-claudebot-robots
- side: claudebot

> Anthropic's Bots respect 'do not crawl' signals by honoring industry standard directives in robots.txt
### claudebot-crawldelay

Anthropic supports the non-standard Crawl-delay extension to robots.txt, shown with ClaudeBot as the example user agent.

- source: https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (support.claude.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-claudebot-crawldelay
- side: claudebot

> To limit crawling activity, we support the non-standard Crawl-delay extension to robots.txt
### claudebot-ips

Anthropic publishes the addresses used by its crawlers as one machine-readable list at claude.com/crawling/bots.json.

- source: https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (support.claude.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-claudebot-ips
- side: claudebot
### claudebot-ua-unpublished

Anthropic's crawler article names its bots by token and does not publish a full user-agent string for ClaudeBot.

- source: https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (support.claude.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-claudebot-ua-unpublished
- side: claudebot
### claudebot-block-effect

Anthropic states that restricting ClaudeBot signals that a site's future materials should be excluded from its AI model training datasets.

- source: https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (support.claude.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-claudebot-block-effect
- side: claudebot

> it signals that the site's future materials should be excluded from our AI model training datasets
### anthropic-agents

Anthropic documents three bots, ClaudeBot, Claude-User and Claude-SearchBot, for model development, user-directed retrieval and search.

- source: https://support.claude.com/en/articles/8896518-does-anthropic-crawl-data-from-the-web-and-how-can-site-owners-block-the-crawler (support.claude.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-anthropic-agents
- side: claudebot
### gptbot-purpose

OpenAI documents GPTBot as used to crawl content that may be used in training its generative AI foundation models.

- source: https://developers.openai.com/api/docs/bots (developers.openai.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-gptbot-purpose
- side: gptbot

> Used to crawl content that may be used in training our generative AI foundation models
### gptbot-token

GPTBot is controlled with the robots.txt user-agent token GPTBot.

- source: https://developers.openai.com/api/docs/bots (developers.openai.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-gptbot-token
- side: gptbot
### gptbot-ua

OpenAI publishes GPTBot's user-agent string as Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.4; +https://openai.com/gptbot

- source: https://developers.openai.com/api/docs/bots (developers.openai.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-gptbot-ua
- side: gptbot
### gptbot-no-crawldelay

OpenAI's crawler documentation does not document Crawl-delay support for GPTBot; it documents robots.txt tokens and published address ranges.

- source: https://developers.openai.com/api/docs/bots (developers.openai.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-gptbot-no-crawldelay
- side: gptbot
### gptbot-ips

OpenAI publishes GPTBot's address ranges in a crawler-specific file at openai.com/gptbot.json.

- source: https://developers.openai.com/api/docs/bots (developers.openai.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-gptbot-ips
- side: gptbot
### gptbot-block-effect

OpenAI states that disallowing GPTBot indicates a site's content should not be used in training generative AI foundation models.

- source: https://developers.openai.com/api/docs/bots (developers.openai.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-gptbot-block-effect
- side: gptbot

> Disallowing GPTBot indicates a site's content should not be used in training generative AI foundation models.
### openai-bot-count

OpenAI's crawler documentation lists GPTBot alongside separate agents for search (OAI-SearchBot) and user-triggered fetches (ChatGPT-User), each with its own robots.txt token.

- source: https://developers.openai.com/api/docs/bots (developers.openai.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-openai-bot-count
- side: gptbot
### study-claudebot-allowed

Of the 8,036 top-10,000 Tranco domains whose robots.txt Attensira read on 3 September 2026, 276 (3.4%) allow ClaudeBot by name.

- source: https://attensira.com/data/ai-crawler-robots-access (attensira.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-study-claudebot-allowed
- side: claudebot
### study-gptbot-allowed

Of the same 8,036 domains, 349 (4.3%) allow GPTBot by name.

- source: https://attensira.com/data/ai-crawler-robots-access (attensira.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-study-gptbot-allowed
- side: gptbot
### study-claudebot-unnamed

Of the same 8,036 domains, 6,683 (83.2%) never name ClaudeBot.

- source: https://attensira.com/data/ai-crawler-robots-access (attensira.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-study-claudebot-unnamed
- side: claudebot
### study-gptbot-unnamed

Of the same 8,036 domains, 6,540 (81.4%) never name GPTBot.

- source: https://attensira.com/data/ai-crawler-robots-access (attensira.com)
- retrieved: 2026-10-04
- cite: https://attensira.com/compare/claudebot-vs-gptbot#claim-study-gptbot-unnamed
- side: gptbot