What is OAI-SearchBot?

How it identifies itself

OAI-SearchBot/1.0; +https://openai.com/searchbot
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; OAI-SearchBot/1.0; +https://openai.com/searchbot

agenttraffic recognises it by the name OAI-SearchBot.

What it does

An AI search crawler builds the index an assistant cites from. It visits ahead of any question, so its hits do not correspond to readers. More on ai search crawler agents.

Is it really them?

OpenAI publishes the IP addresses this agent uses (35 address ranges, list dated 2026-01-02, specific to OAI-SearchBot). agenttraffic checks every OAI-SearchBot request against them and reports it as verified, spoofed, or unchecked when the log has no client IP.

• https://openai.com/searchbot.json: 35 address ranges, published 2026-01-02, specific to this agent. Copy shipped with agenttraffic: 2026-08-28.

Impersonation of OAI-SearchBot is rare but rising, since some scrapers assume search crawlers are allowed where training crawlers are blocked. Verification against OpenAI's published searchbot ranges separates them.

How to block it

In robots.txt it goes by OAI-SearchBot. Copy one of these:

# Allow
User-agent: OAI-SearchBot
Allow: /

# Block
User-agent: OAI-SearchBot
Disallow: /

What it means in your logs

OAI-SearchBot visits look like a classic search engine crawl: a spread of URLs, a fondness for pages linked from your homepage and sitemap, and periodic re-fetches of pages that changed. None of those hits are readers. What they represent is eligibility. A page that OAI-SearchBot has never fetched cannot be cited by ChatGPT search, so the page list under this agent is effectively your ChatGPT index coverage.

agenttraffic classifies it as search and includes it, together with live fetches, in the "pages assistants fetched" section, because both feed what people ultimately see inside an assistant.

Where the verification data comes from

OpenAI publishes a dedicated list of IP addresses for OAI-SearchBot. agenttraffic ships a snapshot with its publish date and checks each request against it, reporting verified, spoofed, or unchecked. Since the same operator publishes three separate lists, a request that matches the GPTBot list while carrying the OAI-SearchBot name is still marked spoofed: verification is per agent, not per operator, when the operator publishes per-agent data.

Blocking OAI-SearchBot

Blocking this robots.txt name removes your site from ChatGPT search results and from the citations ChatGPT shows next to answers. That is the opposite of what most sites want, and it does not stop training, which is GPTBot's job. The usual configuration is to allow OAI-SearchBot and decide about GPTBot separately. OpenAI documents that OAI-SearchBot honours its robots.txt name and that its index respects a GPTBot disallow only for GPTBot, not for search.

Related

The three OpenAI agents form the most searched comparison on this site; see the GPTBot vs ChatGPT-User vs OAI-SearchBot page for a side-by-side table of what each one does and which robots.txt rule affects which product.

Coverage over time

Because OAI-SearchBot re-fetches changed pages, a stable site shows a slow steady trickle and a site with new content shows a burst shortly after publication, followed by re-fetches over the next days. Comparing the page list from one week to the next tells you whether new pages are entering the index promptly. A page that has been live for a fortnight and never appears under this agent is one ChatGPT cannot cite; the fix is the same as for any search engine: link it from a page that is already crawled and list it in your sitemap.

Related agents

• ChatGPT-User (OpenAI, live fetch)
• GPTBot (OpenAI, training)

Published 2026-08-28, updated 2026-08-28. Facts come from the agenttraffic registry.