OpenAI & ChatGPT IP Address List

Live IP ranges for OpenAI crawlers and API egress, fetched directly from OpenAI's official machine-readable JSON feeds and updated automatically.

Crawler IP Ranges

OpenAI bots that crawl your website. Use these ranges to verify incoming requests are genuine OpenAI crawlers, or to configure firewall and CDN rules.

API Egress IP Ranges

Outbound IPs that OpenAI infrastructure uses when calling your endpoints — for example from Agents, Codex Cloud, or ChatGPT integrations. Allowlist these in your firewall to accept traffic from OpenAI-powered systems.

IP Geolocation Finder
Enter any IPv4 or IPv6 address to find its geographic location and network details.

Common Misconceptions

There is no published list of every ChatGPT server IP — only crawler and egress ranges are disclosed.
API traffic does not originate from a fixed, publicly available IP list you should hardcode.
OpenAI publishes crawler IP ranges for GPTBot, SearchBot, and ChatGPT-User.
OpenAI publishes egress IP ranges for ChatGPT Connectors and Agents / Codex Cloud.
You can verify genuine crawlers by cross-referencing the requesting IP against the official ranges above.
Ranges are published as machine-readable JSON feeds — this page fetches them directly so you never need to check manually.

How to Use These IP Ranges

Verifying genuine crawlers

Malicious scrapers frequently spoof the User-Agent of GPTBot to bypass security rules. Cross-referencing the requesting IP against the official crawler ranges above confirms whether a request is genuinely from OpenAI's infrastructure.

Blocking or allowing crawlers via robots.txt

User-agent: GPTBot
Allow: /   # allow training crawl

User-agent: OAI-SearchBot
Allow: /   # allow ChatGPT search indexing

Firewall allowlisting for API egress

If your API endpoint should only accept requests from known OpenAI infrastructure, import the ChatGPT Connectors or Agents / Codex Cloud ranges into your CDN or WAF allowlist. Use the export button in each tab to download as CSV, JSON, or TXT.

Handling ChatGPT-User requests

ChatGPT-User requests are triggered in real time by a human sharing a URL inside ChatGPT. Blocking this range prevents your users from summarising or analysing your content with AI tools — consider carefully before restricting it.

About the data

Prior to 2025, many websites published manually maintained static lists of OpenAI IP ranges. OpenAI now recommends consuming their official machine-readable JSON feeds instead, which update automatically as their infrastructure changes. This page fetches those feeds directly, so the ranges shown are always current.