Skip to content
GP

GPTBot/1.3

UndecidedCommunity-sourced

GPTBot/1.3 is a web crawler developed by OpenAI to assist in gathering information and improving the training of AI models. It uses the capabilities of the GPT (Generative Pre-trained Transformer) architecture to efficiently scrape and analyze content from the internet. This bot aims to enhance the quality and performance of AI by indexing a diverse set of data, ensuring that models stay current with the latest information. The bot operates under specific guidelines to respect website preferences, such as complying with the "robots.txt" file, which indicates the areas a crawler can access. Additionally, GPTBot helps website administrators understand its activity by providing visibility into the types of data being collected. By utilizing this structured approach, OpenAI aims to create more robust and accurate AI systems while maintaining ethical standards in data collection. For more detailed and updated information, you can refer to the official OpenAI page on GPTBot. Operated by OpenAI.

Undecided
Assessment
AI Crawler
Category
Bot
Kind
1
Data sources
AI-generated summary, drawn from Cloudflare, operator docs, and community data

What is GPTBot/1.3?

GPTBot is an automated web crawler developed by OpenAI specifically to gather publicly available web content for the purpose of training its large-scale generative AI models. Unlike search-oriented crawlers, its primary function is to ingest data that informs the reasoning, language capabilities, and general knowledge base of future iterations of the GPT series. OpenAI maintains a clear policy regarding this bot, providing mechanisms for webmasters to restrict its access via robots.txt files.

Helpful — It is a transparent, documented crawler that respects robots.txt directives, allowing site owners to opt-out of having their content used for model training.

Identification

Bot name
GPTBot/1.3
Operator
OpenAI
Category
AI Crawler
Verification
Community-sourced
User-agent string
Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; GPTBot/1.3; +https://openai.com/gptbot)
Main use cases
Training generative AI foundation modelsImproving model accuracy and safetyExpanding the baseline knowledge of AI systems

Why it crawls your site

The bot crawls the web to build a diverse and comprehensive dataset that helps OpenAI's models understand human language, facts, and context. By processing this information, the models become more capable of generating accurate and relevant responses. Site owners can verify the authenticity of the bot by checking the IP ranges published on OpenAI's official documentation page or by inspecting the user-agent string, which includes a link to their crawler overview.

IP ranges

Requests from GPTBot/1.3 originate from these published ranges. For strict verification, match the source IP against this list in addition to checking the user-agent and reverse-DNS.

Showing 24 of 199 ranges

Block or allow GPTBot/1.3

Block it

Add a Disallow rule for Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; GPTBot/1.3; +https://openai.com/gptbot) in your robots.txt file. You can also block at the server level using your web server configuration or CDN firewall rules to filter requests matching the user-agent string.

# Block GPTBot/1.3 from crawling your entire site
User-agent: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; GPTBot/1.3; +https://openai.com/gptbot)
Disallow: /

# Allow GPTBot/1.3 full access
User-agent: Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; GPTBot/1.3; +https://openai.com/gptbot)
Allow: /
Allow & verify

Ensure your robots.txt allows Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko; compatible; GPTBot/1.3; +https://openai.com/gptbot). Verify requests by checking the user-agent string.

Traffic & trends

Cloudflare Radar traffic

No Cloudflare traffic data available.

Google Trends interest

Loading trend data…

Data sources

This profile is compiled from the following sources.

CrawlersDatabase
Last updated September 22, 2026

Something wrong with this entry? Suggest a correction