OAI-SearchBot: definition and how it works

OAI-SearchBot is the user agent OpenAI uses to crawl pages for ChatGPT's search features. OpenAI's crawler documentation says it "is used to surface websites in search results in ChatGPT's search features."

It is one of three OpenAI agents, and each does a different job:

• OAI-SearchBot: search inclusion in ChatGPT.
• GPTBot: crawling for model training only.
• ChatGPT-User: fetches triggered by a user's action, where OpenAI says robots.txt rules "may not apply."

OpenAI states that "each setting is independent of the others." For the full breakdown, see OpenAI's crawler documentation, digested.

As of October 2026, OpenAI's help centre describes eligibility as a conjunction. You must allow OAI-SearchBot to crawl the site and confirm your host or CDN allows traffic from OpenAI's published searchbot IP addresses. OpenAI lists those ranges at openai.com/searchbot.json.

The robots.txt half looks like this:

User-agent: OAI-SearchBot
Allow: /

The network half is the one most sites never test. A firewall or bot-management rule can return a 403 to a crawler that robots.txt allows, and nothing in robots.txt reveals it. OpenAI publishes IP lists but also warns against relying on IP-only allowlisting, so verify by user agent plus your CDN's verified-bot settings. Our guide to testing whether your firewall lets AI crawlers in walks through it.

One more limit: Vercel found that ChatGPT's crawlers fetch JavaScript files but don't execute them. Content that only appears after JavaScript runs may be invisible to OAI-SearchBot.

OAI-SearchBot is an access gate, and access is the stage you control. If the bot can't fetch your page, ChatGPT search can't retrieve it, however good the page is.

What it does not control matters just as much. ChatGPT decides per query whether to search at all, rewrites the prompt into its own queries, and ranks results using what OpenAI calls "multiple factors." Allowing the bot makes you eligible. It does not make you selected or cited.

The practical move: confirm the bot is allowed, confirm your CDN returns a 200 to it, and confirm your key facts are in the raw HTML. Then read the technical checklist for ChatGPT search.

• "Allow GPTBot so ChatGPT can find you." GPTBot governs training only. Search uses OAI-SearchBot. See whether to block GPTBot if you want ChatGPT search traffic.
• "Blocking GPTBot removes you from ChatGPT search." It doesn't. OpenAI says you can allow OAI-SearchBot while disallowing GPTBot.
• "robots.txt is enough." OpenAI's own wording requires network reachability too.
• "Blocking OAI-SearchBot makes you disappear from ChatGPT entirely." Opted-out sites can still appear as navigational links.
• "No visits in my logs means I'm blocked." Crawl patterns differ by bot. See why your logs show GPTBot but not OAI-SearchBot.

• Navigational link: how ChatGPT can still link to a site that opted out of OAI-SearchBot.
• GPTBot: OpenAI's training crawler, set independently.
• ChatGPT-User: OpenAI's user-triggered fetcher.
• robots.txt template for AI search and training bots: a starting file that separates search bots from training bots.

Does allowing OAI-SearchBot guarantee ChatGPT will cite me? No. Allowing the bot makes your pages eligible for ChatGPT search, nothing more. OpenAI says ChatGPT ranks results using multiple factors and that placement is not guaranteed. ChatGPT also decides per query whether to search at all. What you control is access: robots.txt permission, network reachability, and content present in the raw HTML.

What happens if I block OAI-SearchBot? OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as navigational links. Blocking it does not stop GPTBot, which follows its own robots.txt rule, and it may not stop ChatGPT-User, which handles fetches a user triggers.

Can my CDN block OAI-SearchBot even if robots.txt allows it? Yes. Bot-management and firewall rules can return a 403 or a challenge page to automated crawlers. OpenAI's help centre says eligibility requires both robots.txt permission and a host or CDN that allows its published searchbot IP addresses. Test by requesting a page with the OAI-SearchBot user agent and checking the status code.

Does OAI-SearchBot read JavaScript-rendered content? Vercel's crawler study found that ChatGPT's crawlers download JavaScript files but don't execute them, and later independent testing reported the same for OAI-SearchBot. If your prices, product names or key claims load only after JavaScript runs, the crawler may not see them. Server-side rendering of that content removes the gap.