What Is ChatGPT-User?
ChatGPT-User is the OpenAI fetcher triggered on demand when a ChatGPT user query needs live web content. Unlike GPTBot (training crawler running on a schedule) and OAI-SearchBot (search-index crawler refreshing the search corpus), ChatGPT-User runs only when an active user is asking ChatGPT something that requires live browsing.
The result is a distinctive traffic pattern: bursts during peak human ChatGPT usage hours, near-silence overnight, and tight correlation with current news cycles or queries that demand fresh data. ChatGPT-User is the OpenAI crawler identity most directly tied to live citation visibility.
Why ChatGPT-User Matters Disproportionately
For brands competing for live citation in ChatGPT answers (browse-enabled queries, news questions, product comparisons), ChatGPT-User is the moment of truth. The fetch happens because a real human just asked a question that needed your content, and the result of the fetch determines whether your brand or content appears in the live answer the human reads seconds later.
This makes ChatGPT-User fetches several orders of magnitude higher value per fetch than GPTBot fetches, on average. A ChatGPT-User fetch that succeeds and returns relevant content is one citation in one human-facing answer. A GPTBot fetch contributes infinitesimally to the next training cycle in months or years.
Pay-Per-Crawl Behaviour
ChatGPT-User shows the highest 402 compliance among OpenAI crawler identities as of April 2026. When a ChatGPT-User fetch hits a paid content surface and the price is reasonable, the fetch typically pays. The reason is operational: ChatGPT-User has a live human waiting for an answer; walking away from the fetch means an inferior answer for the user. Paying and proceeding is the path of lower friction for the live experience.
For publishers running Pay-Per-Crawl, ChatGPT-User payments are typically the most reliable revenue stream. The volume is moderate (less than GPTBot or OAI-SearchBot in raw fetches), but the pay-rate is the highest of the three.
Performance Considerations
Because ChatGPT-User fetches happen with a human waiting, performance matters. Slow pages return less complete content to ChatGPT-User before the answer is generated, which can cause partial or absent citations even when the fetch succeeded. Pages serious about live citation in ChatGPT browse-enabled answers should target sub-1-second time-to-first-byte and avoid client-side rendering that does not finish before the agent stops waiting.
How to Identify ChatGPT-User in Logs
The user-agent string includes "ChatGPT-User" with a versioned identifier. IP ranges are documented by OpenAI and update periodically. The pattern in logs is unmistakable: bursts during business hours, near-silence overnight, often clustered around specific URLs that align with current human-driven queries. The correlation with downstream citation events is the highest among any AI bot identity.
