ChatGPT-User
OpenAI · User-triggered fetcher
Visits a page when a ChatGPT or Custom GPT user asks about it (also GPT Actions). It does not crawl automatically.
OpenAI says that because these visits are user-initiated, robots.txt rules may not apply.
- User-agent contains
ChatGPT-User- robots.txt token
ChatGPT-User- Respects robots.txt
- Operator docs don't say / may not apply
- Official IP ranges
- 234 prefixes
How to verify ChatGPT-User
Matching the user agent isn't enough, because any client can send it. A real ChatGPT-User request comes from an IP address inside OpenAI's published list:
- https://openai.com/chatgpt-user.json234 prefixes · list dated 2026-10-20 · fetched 2026-10-21
The quickest way to check a whole log is our free log checker. It runs in your browser and splits ChatGPT-User requests into real and fake.
First 20 prefixes
104.208.184.192/28 104.210.139.192/28 104.210.139.224/28 128.85.198.32/28 13.65.138.112/28 13.67.72.16/28 13.71.2.208/28 13.76.116.80/28 13.76.32.208/28 13.83.167.128/28 13.83.237.176/28 132.196.82.48/28 134.138.52.128/28 134.138.52.16/28 134.138.52.64/28 134.138.52.96/28 134.138.57.176/28 134.138.57.64/28 134.138.57.80/28 134.138.58.32/28
Block or allow ChatGPT-User in robots.txt
User-agent: ChatGPT-User Disallow: /
Use Allow: / instead to explicitly permit it, or build a full file with the AI robots.txt generator. robots.txt is a request, not a lock: it only works for bots that honor it, and only real ChatGPT-User traffic would.
Source: OpenAI documentation
Verify ChatGPT-User automatically
One API call tells your app, WAF or middleware whether a request claiming to be ChatGPT-User really comes from OpenAI's official IP ranges.