# Google-Extended vs GPTBot

Google-Extended and GPTBot compared on 10 documented fields, from 16 claims, last verified 2026-09-13.

| Field | Google-Extended | GPTBot | Comparison |
| --- | --- | --- | --- |
| honours_crawl_delay | n/a | not_documented | different |
| publishes_ip_list | no | yes | different |
| purpose_documented | policy_only | training | different |
| respects_robots_txt | n/a | yes | different |
| robots_token_differs_from_ua | yes | not documented | different |
| robots_txt_fetch_marker | not documented | yes | different |
| separate_search_and_training_tokens | yes | yes | same |
| user_agent_string_full | not documented | Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.4; +https://openai.com/gptbot | different |
| verifiable_by_rdns | n/a | not_documented | different |
| what_it_controls | Gemini model training and grounding, not Search inclusion | not documented | different |

## Differences

| Field | Google-Extended statement | GPTBot statement |
| --- | --- | --- |
| honours_crawl_delay | Google does not publish any Crawl-delay behaviour for the Google-Extended token. | OpenAI does not publish whether GPTBot honours the robots.txt Crawl-delay directive. |
| publishes_ip_list | Google publishes no IP list for Google-Extended because Google-Extended has no separate HTTP request user agent string. | OpenAI publishes the IP address ranges used by GPTBot as a JSON file at https://openai.com/gptbot.json. |
| purpose_documented | Google-Extended is a standalone product token that controls whether Google uses crawled content to train Gemini models. | GPTBot crawls content that may be used to train OpenAI generative AI foundation models. |
| respects_robots_txt | Google-Extended is a robots.txt token that publishers set to control use of crawled content for Gemini training. | GPTBot honours robots.txt rules that webmasters set for the GPTBot token. |
| robots_token_differs_from_ua | Google-Extended is a robots.txt control token and does not correspond to a crawler user agent string. | not documented |
| robots_txt_fetch_marker | not documented | OpenAI may add a robots.txt marker to the GPTBot user agent string when GPTBot fetches robots.txt files. |
| user_agent_string_full | not documented | GPTBot sends the user agent string Mozilla/5.0 AppleWebKit/537.36 (KHTML, like Gecko); compatible; GPTBot/1.4; +https://openai.com/gptbot. |
| verifiable_by_rdns | Google does not publish a reverse-DNS verification method for Google-Extended. | OpenAI does not publish a reverse-DNS verification method for GPTBot. |
| what_it_controls | Google-Extended controls AI training and grounding in Google systems and does not control Google Search inclusion. | not documented |

## Sources

- https://developers.google.com/search/docs/appearance/ai-features
- https://developers.google.com/search/docs/crawling-indexing/google-common-crawlers
- https://developers.openai.com/api/docs/bots
- https://openai.com/gptbot.json