OpenAI’s automated mining activity is estimated to have tripled after the launch of GPT-5, according to a new analysis by Botify and guest author Chris Long.
In Botify’s dataset, OpenAI’s search bot now generates more log events than its training bot. This is a reversal from the period before GPT-5.
Long, co-founder of SEO consultancy Nectiv, analyzed approximately 7 billion OpenAI bot log events from Botify’s enterprise customer dataset between November 2024 and March 2026.
What the data shows
Two of the three OpenAI user agents measured by Botify saw an increase in activity around the launch of GPT-5.
OAI-SearchBot, which retrieves content when ChatGPT searches the web, recorded approximately 3.5 times more events after August 2025. This equates to approximately 2.2 billion additional events in Botify’s dataset.
GPTBot, which collects training data, recorded approximately 2.9 times more events over the same period. This represents 1.8 billion additional events.
The third user agent, ChatGPT-User, moved in the opposite direction. Long reports a 28% drop in ChatGPT-User log events between December 2025 and March 2026. ChatGPT-User is triggered when a ChatGPT session fetches a page on behalf of a user, so the drop measures user-initiated fetches rather than overall ChatGPT usage.
Long offers two possible readings. The first is that fewer sessions can trigger real-time page fetches. The other, suggested by the Botify team, is that OpenAI could rely more on stored or indexed resources, reducing the need to fetch pages in real time. Long doesn’t choose between them.
The research robot now exceeds the training robot
Prior to GPT-5, OAI-SearchBot and GPTBot performed at roughly equal volumes across the Botify dataset, with a ratio of approximately 0.95 search events per training event. After GPT-5, this ratio increased to around 1.14.
The model matches what Dan Petrovic wrote in August 2025 about GPT-5arguing that OpenAI obtained more answers from live search than from trained memory. Botify’s data is consistent with this reading.
Industry breakdown
The increase in search bots post-GPT-5 varies by industry. Healthcare sites saw approximately 740% more OAI-SearchBot activity after launch; Media and publishing, 702%; and Markets, Software and Retail, 190-216%.
Travel sites saw the smallest increase at 30%. The balance between research and training also varies. Long reports an OAI-SearchBot versus GPTBot analysis difference of +256% for Media/Publishing, the largest gap. Software and Internet lean towards search, Health and Retail favor training, with -50% and -33%. GPTBot is generally more active.
Botify and Long suggest that OpenAI routes prompt types differently: information requests trigger live searches, health and product queries rely on qualified knowledge.
How OpenAI’s analytics compare to Google’s
Even after tripling, OpenAI’s crawl activity is much lower than Google’s.
During Botify’s most recent 30-day window, Googlebot logged 18.2 billion events, compared to 887 million events for OpenAI’s crawlers combined. This puts OpenAI at around 4% of Google’s crawl volume.
A year earlier, the same comparison was 15 billion Google events versus 207 million OpenAI events, or about 1.38%. The gap is narrowing, although Google’s crawl is still around 20 times larger in absolute terms.
Bingbot logged about 5.49 billion events during the most recent window, putting OpenAI at about 14% of Bing.
Methodology & Commercial Context
The dataset is owned by Botify and covers enterprise customers across the retail, e-commerce, technology, publishing, travel, and marketplace industries. The analysis was conducted by Long as a guest author on the Botify blog.
For the sake of transparency, Botify sells software for log file analysis and AI bot management, and the post promotes a follow-up webinar and product demo.
The dataset is skewed toward large company websites rather than a representative sample of the web.
Why it matters
In the Botify dataset, OAI-SearchBot now generates more log events than GPTBot. Sites that only block GPTBot do not block the bot which OpenAI says is used to make websites appear in ChatGPT search responses.
Sites that block OAI-SearchBot may exclude themselves from ChatGPT search responses.
How this fits with other reports
Botify’s results are consistent with patterns reported by other vendors. An Alli AI analysis covered earlier this month revealed OpenAI user ChatGPT made 3.6 times more queries than Googlebot in a smaller, heavier sample of WordPress. Hostinger analysis found OAI-SearchBot website coverage reaches 55% while GPTBot coverage dropped. Akamai’s recent bot traffic report showed OpenAI leader in AI bot traffic to publishing sites.
Reports suggest that AI training and search explorations should be measured separately, particularly as OAI-SearchBot activity grows.
