AI Bot Traffic Is Accelerating Fast. 48 Days of Server Logs Expose What GPTBot, ChatGPT, ClaudeBot, and 16 Others Are Doing.
Published: March 20, 2026
AI Bot Traffic Is No Longer a Trickle
Between February and mid-March 2026, AI bot activity on wislr.com didn’t just grow. It changed shape.
- ChatGPT-User traffic 5x’d in seven weeks
- GPTBot, completely absent until March 12, executed 187 requests in a single week, 152 of them in a 3-minute burst
- GPTBot is not new. It is OpenAI’s core training crawler, active across the web for years. But it does not appear to crawl every site continuously. The data suggests it activates on a site once that site’s content gains traction in OpenAI’s ecosystem. ChatGPT-User referrals to wislr.com had already 5x’d before GPTBot ever showed up.
- OAI-SearchBot volume 4x’d
- On March 18-19, both ClaudeBot and GPTBot started requesting sitemap.xml for the first time, on the same day, from different companies
That is a shift in how AI platforms discover and index content.
20 Takeaways for Your Own Website
- Google Analytics cannot see any of this. AI bots do not execute JavaScript. If you rely on client-side analytics, your AI bot traffic is invisible. Server-side logging is the only way to measure it.
- Your sitemap.xml just became more important. GPTBot and ClaudeBot both started consuming sitemaps in March 2026 for the first time. If your sitemap is stale, incomplete, or missing language variants, AI crawlers will miss content.
- robots.txt is not universally respected. GPTBot and Meta-WebIndexer never check it. If your AI content strategy depends on robots.txt directives, know that two of the most active crawlers ignore them entirely.
- Multilingual content gets disproportionate crawl attention. Bots like Meta-WebIndexer (80%), GPTBot (62%), and Bingbot (60%) spend the majority of their budget on language variants. If you publish translated content, AI platforms are indexing it aggressively.
- ChatGPT-User traffic is a direct signal of brand citation in AI conversations. Each request represents a real person pasting your URL into ChatGPT. This is measurable word-of-mouth, and it is growing fast.
- AI bots crawl in bursts, not steady streams. GPTBot hit 114 req/min in a 3-minute window. If your server can’t handle burst traffic, AI crawlers may get throttled or hit errors during their indexing runs.
- OpenAI and Anthropic each operate 3 separate bots. One for training/indexing, one for search, one for live user sessions. Blocking one does not block the others. Your robots.txt needs separate directives for each.
- OAI-SearchBot and Googlebot are the only bots that fetch images at volume. If your article images carry meaningful content (charts, diagrams, data visualizations), these are the bots that will use them in search results.
- ChatGPT-User only extracts text. Zero images, zero CSS, zero JS. Your HTML content is what gets pulled into AI conversations. Structured, clear text matters more than visual design for AI visibility.
- AI crawlers peak at different hours. GPTBot hits at 04:00 UTC. Claude-SearchBot peaks overnight. PerplexityBot bursts at 23:00, 05:00, and 09:00. If you deploy site changes during off-peak US hours, AI bots may be the first to see them.
- Meta is the most aggressive AI crawler by volume. Meta-WebIndexer sent more requests than any other bot in this dataset, with zero robots.txt checks. If you are not tracking Meta’s crawlers, you are missing the biggest player.
- llms.txt adoption is still theoretical. Zero AI bots requested /llms.txt across 48 days. It may become a standard eventually, but no crawler currently looks for it.
- Applebot renders your pages fully. It fetches CSS, JS, and images (47% of its traffic). If your content requires JavaScript rendering to be complete, Applebot will see it, but most AI bots will not.
- ChatGPT-User traffic is globally distributed. 15 countries, 584 unique IPs. Your content is being referenced in AI conversations worldwide, not just in the US.
- Technical, how-to content gets referenced most in AI conversations. The top ChatGPT-User pages were all implementation guides and technical explainers. Deep, specific content earns AI citations.
- Bytespider and CCBot only check robots.txt and never crawl. They are consuming your robots.txt directives without following through. This may change, but currently they generate compliance overhead with zero content indexing.
- AI crawl volume can shift overnight. GPTBot went from 0 to 187 requests in a single week. Your crawl budget projections need to account for sudden step-changes, not gradual growth.
- IP analysis reveals bot identity. ChatGPT-User’s near 1:1 IP-to-request ratio proves individual user sessions. GPTBot’s 2 IPs prove centralized infrastructure. IP patterns help distinguish real user-triggered fetches from automated crawling.
- Coordinated crawl events happen across bot families. GPTBot and OAI-SearchBot fired simultaneously on March 19 from the same Microsoft infrastructure. When one OpenAI bot ramps up, expect the others to follow.
- The bots you have never heard of are already visiting. PromptingBot, LinkupBot, Brightbot, Observer, and others are actively crawling content. The AI bot landscape is larger than the well-known names suggest.
The Data
Source: 288,566 log files from CDN script request logging on wislr.com
Period: February 1 through March 20, 2026
Total requests: 71,603
AI/crawler bot requests analyzed: 12,099 (16.9% of all traffic)
Every HTTP request to wislr.com passes through a CDN script that captures full request metadata (user agent, geo, headers, timing). This gives us ground truth on exactly what AI bots are doing, how often, and what they are reading.
Want this level of visibility into your own AI bot traffic?
Schedule a ChatBot Volume Rankings
| Rank | Bot | Total Hits | Feb | Mar (partial) | Trend |
|---|---|---|---|---|---|
| 1 | Meta-WebIndexer | 1,833 | 707 | 1,126 | +59% |
| 2 | Bingbot | 1,221 | 547 | 674 | +23% |
| 3 | ChatGPT-User | 923 | 230 | 693 | +201% |
| 4 | Googlebot | 816 | 386 | 430 | +11% |
| 5 | Claude-SearchBot | 549 | 216 | 333 | +54% |
| 6 | Applebot | 495 | 257 | 238 | -7% |
| 7 | PerplexityBot | 456 | 244 | 212 | -13% |
| 8 | OAI-SearchBot | 330 | 117 | 213 | +82% |
| 9 | ClaudeBot | 206 | 69 | 137 | +99% |
| 10 | GPTBot* | 187 | 0 | 187 | New in March |
| 11 | Amazonbot | 71 | 27 | 44 | +63% |
| 12 | Bytespider | 68 | 33 | 35 | +6% |
| 13 | Claude-User | 18 | 6 | 12 | +100% |
| 14 | CCBot | 5 | 2 | 3 | Minimal |
| 15 | Meta-ExternalAgent | 4 | 0 | 4 | New |
| 16 | DuckAssistBot | 4 | 0 | 4 | New |
| 17 | Perplexity-User | 3 | 3 | 0 | Minimal |
| 18 | YouBot | 2 | 0 | 2 | New |
| 19 | GoogleOther | 2 | 1 | 1 | Minimal |
*GPTBot is not a new bot. It is OpenAI’s core training crawler, active across the web for years. “New in March” means new to this site. The data suggests GPTBot activates on a site once its content gains traction in OpenAI’s ecosystem: ChatGPT-User referrals had already 5x’d before GPTBot ever appeared.
Weekly Volume Trends
| Bot | Feb W1 | Feb W2 | Feb W3 | Feb W4 | Mar W1 | Mar W2 | Mar W3 |
|---|---|---|---|---|---|---|---|
| GPTBot | 0 | 0 | 0 | 0 | 0 | 4 | 183 |
| ChatGPT-User | 39 | 66 | 65 | 60 | 72 | 304 | 317 |
| OAI-SearchBot | 28 | 34 | 29 | 26 | 28 | 46 | 139 |
| ClaudeBot | 10 | 14 | 9 | 36 | 51 | 24 | 62 |
| Claude-SearchBot | 31 | 51 | 47 | 87 | 121 | 68 | 144 |
| PerplexityBot | 22 | 132 | 62 | 28 | 37 | 77 | 98 |
| Bingbot | 132 | 163 | 136 | 116 | 151 | 248 | 275 |
| Meta-WebIndexer | 63 | 35 | 271 | 338 | 396 | 541 | 189 |
| Amazonbot | 8 | 6 | 7 | 6 | 2 | 2 | 40 |
March W3 (Mar 15-20) was a breakout week. GPTBot exploded from near-zero to 183 requests. ChatGPT-User sustained 300+/week (5x its February baseline). OAI-SearchBot hit 139, a 4x jump. Something coordinated happened across OpenAI’s bot fleet.
Individual Bot Profiles
GPTBot (OpenAI’s training/indexing crawler)
Total: 187 requests | Active since: March 12 | IPs: 2 (highly concentrated)
GPTBot is not a new bot. OpenAI defines it as the crawler used to “make our generative AI foundation models more useful and safe,” crawling content that may be used in training their foundation models. It has been operating across the web for years. What is notable here is that GPTBot was completely absent from wislr.com until March 12, then activated with a massive crawl burst on March 19 at 04:38 UTC: 152 requests in 3 minutes (114 req/min peak). It systematically crawled every article in every language variant.
This suggests GPTBot does not crawl every site continuously. It appears to activate and ramp up on a site once that site’s content gains traction across OpenAI’s ecosystem, as ChatGPT-User referrals to wislr.com were already 5x’d before GPTBot showed up.
What it crawls:
- Articles: 67% of requests
- Language variants: 61.5%, crawled all 11 language versions (de, fr, pl, hi, nl, uk, pt, ms, tr, it, es)
- Images: 28 requests (article headers, author photos)
- Sitemap.xml: 2 requests (started Mar 19, brand new behavior)
- robots.txt: Zero. Never checks it.
Crawl timing: 81% of all requests at 04:00 UTC. Extreme burst pattern, nearly all traffic in a single 3-minute window.
Infrastructure: Runs from Microsoft Limited ASN, US only. 97.9% of traffic from a single IP.
GPTBot’s burst on March 19 was concurrent with an OAI-SearchBot burst from the same Microsoft infrastructure: 50 req/min at the same timestamp. This appears coordinated.
OAI-SearchBot (OpenAI’s search crawler)
Total: 330 requests | Trend: +82% Feb to Mar | IPs: 81
OAI-SearchBot is OpenAI’s dedicated search crawler, distinct from GPTBot. It is the most persistent robots.txt checker of any AI bot, hitting it 3-6 times every single day without exception.
What it crawls:
- robots.txt: 54.5% of all traffic (180 hits, 3.8/day average)
- Articles: 52 requests
- Images: 55 requests, actively downloads article images, author photos, page screenshots
- Homepage: 9 requests
Crawl timing: Spread across the day with slight peaks at 04:00 and 20:00 UTC.
Infrastructure: Microsoft Limited (87%) + Microsoft Corp (13%), US only.
The only OpenAI bot that respects robots.txt (checks it obsessively). Also the only AI bot besides Googlebot that actively fetches images at volume, likely for search result cards.
ChatGPT-User (live user session fetcher)
Total: 923 requests | Trend: +201% Feb to Mar | IPs: 584
This is not a crawler. It fires when a real person pastes a wislr.com URL into a ChatGPT conversation. The 584 unique IPs (nearly 1:1 with requests) confirm these are individual user sessions. This is the clearest signal of how often wislr.com content is being cited in AI conversations.
Most-referenced content:
| Hits | Article |
|---|---|
| 225 | Shopify Same-Domain Checkout Analytics |
| 112 | Cloudflare Crawl Endpoint Pros and Cons |
| 103 | Cloudflare Crawl Best Settings |
| 91 | Cloudflare CDN Request Logging for Shopify |
| 81 | 6 Core Tasks for Better 301 Redirects |
| 50 | AI Performance Metrics: Seven KPIs |
| 37 | Homepage |
| 30 | Essential Tools for Site Migrations |
| 29 | OpenAI KPIs and Success Metrics 2026 |
| 23 | AEO Readiness Comparison |
What it does not fetch: Zero images, zero CSS, zero JS. Pure HTML content extraction, it only wants the text.
Crawl timing: Evenly distributed across all 24 hours, follows human usage patterns.
Geographic distribution: US (50%), Poland (16%), Australia (5%), Spain (5%), Japan (5%), Brazil (4%), 15 countries total. Top cities: Warsaw (149), Phoenix (145), San Antonio (126), Des Moines (91).
ClaudeBot (Anthropic’s training/indexing crawler)
Total: 206 requests | Trend: +99% Feb to Mar | IPs: 22
ClaudeBot is overwhelmingly focused on compliance checking: 85% of its traffic is robots.txt.
What it crawls:
- robots.txt: 175 requests (85%), increased from 1-2/day in early Feb to 6-12/day by March
- Sitemap.xml: 14 requests, started March 18 (never touched it before)
- Images: 10 requests (all on March 18)
- Content pages: very few
ClaudeBot checks robots.txt more aggressively over time. Its sudden interest in sitemap.xml on March 18 (5 hits that day, 8 on March 19) suggests a new crawl strategy rollout from Anthropic.
Infrastructure: Anthropic, PBC ASN (97%), US only.
Claude-SearchBot (Anthropic’s web search crawler)
Total: 549 requests | Trend: +54% Feb to Mar | IPs: 8
A separate Anthropic bot distinct from ClaudeBot and Claude-User. This powers Claude’s web search feature. More aggressive than ClaudeBot, actually crawls content.
What it crawls:
- robots.txt: 164 requests (30%), daily, increasing from 1-3 to 7-10/day
- Sitemap.xml: 135 requests, started Feb 25, quickly became the second most active sitemap consumer after Bingbot
- Articles: 142 requests (26%)
- Language variants: 23.3% of traffic
- Homepage: 45 requests
Crawl timing: Overnight-weighted, peaks at 22:00, 01:00, 00:00 UTC.
Combined, Anthropic operates 3 distinct bots (ClaudeBot, Claude-SearchBot, Claude-User) totaling 773 requests, making Anthropic the third largest AI bot presence after Meta-WebIndexer and ChatGPT-User.
Claude-User (Anthropic’s live session fetcher)
Total: 18 requests | IPs: 8
The Anthropic equivalent of ChatGPT-User. Fires when someone shares a URL in a Claude conversation. Much lower volume than ChatGPT-User (18 vs 923).
What it crawls: robots.txt (9), then articles about Cloudflare and a law firm case study.
Infrastructure: Google LLC ASN (100%), US only. Runs through Google Cloud, not Anthropic’s own infra.
PerplexityBot (Perplexity’s indexing crawler)
Total: 456 requests | Trend: -13% Feb to Mar | IPs: 13
Bursty crawl pattern: 132 requests in Feb W2, drops to 28 in Feb W4, back up to 98 in Mar W3.
What it crawls:
- Articles: 77% of traffic, heavily targets AI performance metrics content
- robots.txt: 71 requests (steady 1-4/day)
- Images: 18 requests (article-specific images)
- Sitemap.xml: Zero. Never requests it.
- Language variants: 52.9%, broad multilingual coverage
Crawl timing: Three distinct burst windows at 23:00, 05:00, and 09:00 UTC.
Infrastructure: Amazon Technologies Inc. (96%), US only. 13 IPs, moderate concentration.
Googlebot
Total: 816 requests | IPs: 72
Homepage-heavy (27% of traffic) and the most “full render” crawler. Fetches images, CSS, JS alongside content.
What it crawls:
- Homepage: 220 requests (27%)
- Images: 152 requests
- CSS: 132 requests
- JS: 54 requests
- Articles: 141 requests
- robots.txt: 68 requests
- Sitemap.xml: 5 requests (occasional)
Language variants: Only 10.4%, significantly less multilingual than other bots.
Infrastructure: Google LLC (91%), some from Sweden (8%). Uniform 24-hour crawl pattern.
Bingbot
Total: 1,221 requests | Trend: +23% Feb to Mar | IPs: 258
The most consistent and broad crawler. Also the dominant sitemap consumer.
What it crawls:
- Articles: 68% (with heavy language variant coverage)
- Sitemap.xml: 139 requests (2-8/day, every day), by far the most active sitemap crawler
- Images: 78 requests (social icons, dashboard PNGs, article images)
- robots.txt: 36 requests
Language variants: 60.3%, heavy focus on Dutch (97) and German (98) variants.
Crawl timing: Uniform 24-hour coverage, always-on.
Meta-WebIndexer
Total: 1,833 requests | Trend: +59% Feb to Mar | IPs: 156
The highest-volume bot overall. Overwhelmingly focused on content, especially language variants.
What it crawls:
- Articles: 88% of traffic
- Language variants: 79.8%, the most aggressive multilingual crawler by far
- Sitemap.xml: 30 requests
- robots.txt: Zero. Does not check robots.txt at all.
Crawl timing: Evening/overnight heavy, peaks at 20:00-21:00 and 00:00-01:00 UTC.
Infrastructure: Meta Platforms Ireland Ltd ASN, US only.
Like GPTBot, Meta-WebIndexer does not check robots.txt. Unlike GPTBot, it has been consistently active throughout the entire period.
Behavioral Patterns
robots.txt Compliance
| Behavior | Bots |
|---|---|
| Obsessive checkers (multiple times daily) | OAI-SearchBot (3.8/day), ClaudeBot (4/day), Claude-SearchBot (4/day) |
| Regular checkers (daily or near-daily) | PerplexityBot, Googlebot, Applebot, Amazonbot, FacebookExternalHit (exactly 1/day like clockwork) |
| Only checks robots.txt, nothing else | Bytespider (100%), CCBot (100%) |
| Never checks robots.txt | GPTBot, Meta-WebIndexer, ChatGPT-User |
GPTBot and Meta-WebIndexer are the notable non-compliant bots. They crawl aggressively without ever checking the rules.
Sitemap Discovery
| Bot | Sitemap Hits | First Request | Pattern |
|---|---|---|---|
| Bingbot | 139 | Feb 1 | Daily, 2-8/day (the sitemap king) |
| Claude-SearchBot | 135 | Feb 25 | Heavy, became 2nd most active |
| Meta-WebIndexer | 30 | Feb 5 | Regular |
| ClaudeBot | 14 | Mar 18 | Brand new behavior |
| Googlebot | 5 | Feb 2 | Occasional |
| GPTBot | 2 | Mar 19 | Brand new behavior |
| Applebot | 1 | Mar 17 | Single hit |
March 18-19 marks a shift. Both ClaudeBot and GPTBot started requesting sitemap.xml for the first time, suggesting new crawl strategy rollouts from Anthropic and OpenAI simultaneously.
llms.txt / AI-Specific File Requests
No AI bot requests /llms.txt or /llm.txt. Zero hits from GPTBot, ClaudeBot, PerplexityBot, or any other AI crawler.
The only bot checking for llms.txt is Dataprovider.com (a web analytics company): 3 requests across Feb-Mar.
Two requests for /.well-known/ai-plugin.json came from Firefox browsers, not bots.
Language Variant Crawling
| Bot | % of Traffic in Language Variants | Coverage |
|---|---|---|
| Meta-WebIndexer | 79.8% | All 11 languages, evenly distributed |
| GPTBot | 61.5% | All 11 languages, 8-12 each |
| Bingbot | 60.3% | All 11, heavy on Dutch/German |
| PerplexityBot | 52.9% | All 11, German/Spanish top |
| Amazonbot | 32.4% | All 11 |
| Claude-SearchBot | 23.3% | All 11, fairly even |
| Applebot | 23.0% | All 11 |
| OAI-SearchBot | 11.2% | All 11 |
| Googlebot | 10.4% | Spanish/Portuguese top |
| ChatGPT-User | 4.7% | French/Spanish/Italian (user-driven) |
Meta-WebIndexer and GPTBot are the most aggressive multilingual crawlers. ChatGPT-User’s low percentage makes sense: it reflects what languages real users are reading in.
Image Crawling
| Bot | Image Requests | What They Fetch |
|---|---|---|
| Googlebot | 152 | SVG icons, logos, favicons, rendering assets |
| Google-InspectionTool | 108 | Social icons, case study images |
| Googlebot-Image | 75 | Favicon, logo variants, article images |
| Bingbot | 78 | Social icons, dashboard PNGs, article images |
| OAI-SearchBot | 55 | Article images, author photos, screenshots |
| Applebot | 32 | Article cover images |
| GPTBot | 28 | Article headers, author photos (all on Mar 19) |
| PerplexityBot | 18 | Article images |
| ClaudeBot | 10 | Author photo, screenshots (all on Mar 18) |
| ChatGPT-User | 0 | Never fetches images |
Googlebot fetches images for rendering. OAI-SearchBot likely fetches them for search result cards. GPTBot and ClaudeBot both did concentrated image bursts on a single day each, suggesting batch indexing runs.
ChatGPT-User fetches zero images. Pure text extraction.
Other Notable Bots Discovered
PromptingBot/1.0.0 (104 requests)
- Runs on Google Cloud infrastructure
- Active Feb 24 through Mar 13
- Does full page rendering (fetches CSS, JS, images, social icons)
- Targets AI performance metrics content
LinkupBot/1.0 (47 requests)
- From linkup.so, a web indexing service
- Microsoft infrastructure, origin: France
- Checks robots.txt, then focuses on migration/redirect content
Brightbot 1.0 (46 requests)
- From truview LLC, 41 unique IPs
- Focuses on Dutch language variants
Observer/1.0 (36 requests)
- From obsrvr.net, German/Finnish infrastructure
- Only crawls robots.txt and a single French article
Dataprovider.com (148 requests)
- Canadian web analytics company
- The only bot that checks for /llms.txt (3 times)
- Also checks /ads.txt and /humans.txt
HTTP Methods
Every single bot uses GET exclusively. Zero HEAD, POST, or other methods across all 12,099 bot requests.
Country/ASN Data
| Bot | Primary ASN | Countries |
|---|---|---|
| GPTBot | Microsoft Limited | US only |
| OAI-SearchBot | Microsoft Limited (87%) + Microsoft Corp (13%) | US only |
| ChatGPT-User | Microsoft Corporation (78%) | US (50%), Poland (16%), Australia (5%), Spain (5%), Japan (5%), Brazil (4%), 15 countries |
| ClaudeBot | Anthropic, PBC (97%) | US only |
| Claude-SearchBot | Anthropic, PBC (100%) | US only |
| Claude-User | Google LLC (100%) | US only |
| PerplexityBot | Amazon Technologies Inc. (96%) | US only |
| Amazonbot | Amazon (100%) | US only |
| Applebot | Apple Inc. (100%) | US only |
| Bingbot | Microsoft Corporation (99.7%) | US only |
| Googlebot | Google LLC (91%) | US (92%), Sweden (8%) |
| Meta-WebIndexer | Meta Platforms Ireland Ltd (100%) | US only |
| Bytespider | Amazon Singapore (53%), China Unicom (47%) | Singapore (53%), China (47%) |
All major AI bots run from US infrastructure except Bytespider (Singapore/China). ChatGPT-User is the only bot with true global distribution, reflecting its human user base.
Frequently Asked Questions
Which AI bots are visiting websites most frequently in 2026?
Based on 48 days of server log data from wislr.com, the highest-volume AI bots are Meta-WebIndexer (1,833 requests), ChatGPT-User (923 requests), Claude-SearchBot (549 requests), Applebot (495 requests), PerplexityBot (456 requests), OAI-SearchBot (330 requests), ClaudeBot (206 requests), and GPTBot (187 requests). ChatGPT-User showed the sharpest growth at +201% between February and March 2026.
Do AI bots check robots.txt before crawling?
It varies widely. OAI-SearchBot checks robots.txt 3-6 times per day without exception. ClaudeBot and Claude-SearchBot check 4 times per day on average. Bytespider and CCBot only check robots.txt and never crawl any actual content. However, GPTBot and Meta-WebIndexer never check robots.txt at all, despite being among the most active crawlers. ChatGPT-User also never checks robots.txt.
Does any AI bot request llms.txt?
No. Across 12,099 bot requests over 48 days, zero AI bots requested /llms.txt or /llm.txt. The only bot that checked for llms.txt was Dataprovider.com, a web analytics company, not an AI platform. Despite growing discussion around llms.txt as a standard, no AI crawler currently looks for it in practice.
What is the difference between GPTBot and ChatGPT-User?
GPTBot is OpenAI’s training and indexing crawler that systematically crawls pages, language variants, and images in concentrated bursts, often without checking robots.txt. ChatGPT-User fires when a real person pastes a URL into a ChatGPT conversation. ChatGPT-User had 584 unique IPs for 923 requests (nearly 1:1), confirming individual user sessions. ChatGPT-User only fetches HTML text, never images, CSS, or JS. GPTBot fetches all asset types.
How can you detect AI bot traffic on your website?
AI bot traffic is invisible to client-side analytics tools like Google Analytics because bots do not execute JavaScript. You need server-side request logging that captures the raw HTTP requests hitting your domain, including the User-Agent header that identifies bots like GPTBot, ChatGPT-User, ClaudeBot, and others. The WISLR AI Visibility Dashboard provides this level of visibility out of the box.
Which AI bots crawl multilingual content most aggressively?
Meta-WebIndexer spends 79.8% of its crawl budget on language variants, making it the most aggressive multilingual crawler. GPTBot follows at 61.5%, crawling all 11 language versions of articles. Bingbot (60.3%) and PerplexityBot (52.9%) also heavily target multilingual content. Googlebot is notably less multilingual at only 10.4%. ChatGPT-User’s low 4.7% rate reflects that it is driven by human users, not automated crawling.