Pay-per-crawl / log checker
Which AI crawlers are in your server logs?
Paste or upload an access log. You'll see every AI crawler that visited, how many pages it took, whether it really comes from the company it names, and AI company addresses posing as ordinary browsers. Your log never leaves this browser: the check runs here, and nothing is uploaded.
Is this really GPTBot? Check one address
Enter the address a request came from and the user agent it sent (or just a crawler name, such as GPTBot or ClaudeBot). The check runs in this browser against the same published address lists as the log check: checking sends the address nowhere, not even to us.
How the check works
- Named crawlers. A line counts as an AI crawler when its user agent names one of the 27 crawlers in our directory (GPTBot, ClaudeBot, PerplexityBot, CCBot, Bytespider and more).
- Real or fake. OpenAI, Anthropic, Perplexity, Mistral, DuckDuckGo and Common Crawl publish the addresses their crawlers use. A visit that names one of those crawlers from any other address is borrowing the name. Operators that publish no list (ByteDance, Meta and others) can't be checked by address.
- Disguised. A visit from one of those published addresses that sends a normal browser name is an AI company's traffic that doesn't say so.
- The address lists come from /v1/ai_bot_ips, refreshed daily from the operators. Check a single address above, in this browser.