What GPTBot sees
What GPTBot sees is a free viewer that requests one page as OpenAI's GPTBot crawler, without JavaScript, and shows the title, headings, word count and text that arrive, next to the same request from a browser user agent.
Two requests to the page and one to robots.txt. Nothing is stored.
Check every AI crawler and every page
This viewer checks one page as one crawler. The free scan requests up to 10 pages as GPTBot, ClaudeBot, PerplexityBot and CCBot and compares each one with a real browser render.
Read more: why crawlers can't see JavaScript sites and GPTBot, OAI-SearchBot and ChatGPT-User.
Questions
How does this viewer work?
It requests your page once with the GPTBot user agent and once with a normal browser user agent. Neither request runs JavaScript, which is how GPTBot fetches pages. It also reads your robots.txt rule for GPTBot.
Why does GPTBot see an empty page?
Sites built as single-page apps (for example many Lovable, Bolt, v0 or Base44 sites) send a nearly empty HTML file and build the content in the browser with JavaScript. GPTBot does not run JavaScript, so it gets only the empty shell.
Is this exactly what OpenAI receives?
It is close but not exact. Our request uses the GPTBot user agent but comes from our servers, not from OpenAI's IP ranges. A firewall that checks IP addresses can treat the real GPTBot differently.
Is GPTBot the crawler for ChatGPT search?
No. OpenAI uses GPTBot for training, OAI-SearchBot for ChatGPT search results, and ChatGPT-User when a person asks ChatGPT to open a page. All three fetch HTML without running JavaScript, so this view shows the problem for each of them.