Last updated: August 30, 2026
OpenAI started testing ads inside ChatGPT in the United States in February 2026. Eight more markets followed over the next six months. On August 24, 2026, ChatGPT Ads expanded to 31 European countries, including Spain, Germany, France, Italy, the Netherlands and Nordic markets.
Table of Contents
The expansion gives advertisers a much wider market to evaluate.
Until today, there was no way to see what was running inside it.
Every major ad channel has a public archive. This one did not.
Meta has the Ad Library. Google has the Ads Transparency Center. TikTok, LinkedIn and X all publish some version of an ad archive. You can look up a competitor, see their live creatives, see which markets they are running in, and work out how long a message has been in market before you commit budget to your own.
For ChatGPT Ads, that layer did not exist. OpenAI does not publish a searchable public archive, and it has not said whether it will.
If you wanted to know who was advertising inside ChatGPT, you had two options: run prompts by hand and screenshot whatever came back, or ask around. Neither gives you history. Neither tells you what ran last month in a country you are about to enter.
So we built the archive.
What our ChatGPT Ads Library does
Our ChatGPT Ads Library is a public database of ads we detect inside ChatGPT in the countries where the advertising platform is live. Visitors can browse a sample, and signing in adds search and filters.
After signing in, you can:
- Search every ad we have detected, by advertiser or by the words in the message
- Filter by country, which matters more every week as the channel expands
- Filter by date, so you can see when a campaign started and whether it is still running
- Filter by advertiser, to pull a competitor’s full creative history in one view
New observations appear as LLM Pulse detects them. A public sample is free to browse, and signing in unlocks search and filters.
The library currently covers more than 72,000 unique observed ads across 40 enabled countries.
Three things you can do with it today
- Watch your category before you spend. If you are deciding whether ChatGPT Ads deserves a line in your 2026 plan, the useful question is not “what does OpenAI say about the format”, it is “is anyone in my category already buying it, and what are they saying”. The library answers that in about thirty seconds.
- Read the creative, not the spec sheet. ChatGPT ads are not search ads and they are not display. The format is young enough that nobody has settled on what good looks like. Reading real placements is worth more than any best practices post, including this one.
- Time your market entry. With 31 countries opening on the same day, most advertisers will move at the same time. Filtering by country and date shows you who moved first in each market, and how quickly a category filled up.
How we detect the ads, and what that means
We would rather tell you the limits up front than have you find them later.
LLM Pulse tracks AI answers across supported countries and languages on the selected weekly or daily execution cycle. The library is built from the paid placements that show up in those runs. That has two consequences worth stating plainly:
- It is a sample, not a census. An advertiser missing from the library is not proof they are not running. It means we have not caught them yet. Coverage grows as the number of tracked prompts grows.
- Everything in it is real. We are not modelling or estimating what might be running. Every ad in the library is a placement that actually appeared in a ChatGPT answer, with the date it appeared.
This is also not OpenAI data. It is observed data, collected independently. If OpenAI ships an official archive tomorrow, we will happily point you at it.
Why we are giving it away
We are a bootstrapped team of three. We do this because free tools are how most people meet us, and because the alternative was that this data simply did not exist for anyone.
It is the same reason we run Tremor, our free weekly AI Search Volatility Tracker. If the ads library turns out to be useful to you, the paid product is where you track your own brand across ChatGPT, Gemini, Google AI Mode, Google AI Overviews and Perplexity, starting at €49 a month with a 14 day trial.
A public sample of the ChatGPT Ads Library is free to browse.
Take a look
Our ChatGPT Ads Library is live now. Browse it, sign in to search it, and tell us what is missing. We ship fast, and feedback from this post has a decent chance of being a feature by the end of the week.
This is an independent ChatGPT Ads Library built from placements observed by LLM Pulse.
Great work by the LLM Pulse team! 👏
FAQ
What is the ChatGPT Ads Library?
It is a public database of ads observed inside ChatGPT in the countries where the advertising platform is live. Visitors can browse a sample, while signing in unlocks search, filters, ad creatives, and 30 results. The library currently covers more than 72,000 unique observed ads across 40 enabled countries, with new observations added as LLM Pulse detects them.
Is this official OpenAI data?
No. OpenAI does not publish a searchable public ad archive, and it has not said whether it will. Everything in our library is observed data collected independently through LLM Pulse tracking runs, and if OpenAI ships an official archive we will happily point you at it.
How are the ads detected?
LLM Pulse tracks AI answers across supported countries and languages on the selected weekly or daily execution cycle, and the library is built from the paid placements that appear in those runs. Every ad in it is a real placement that actually showed up in a ChatGPT answer, with the date it appeared. Nothing is modelled or estimated.
Does the library show every ad running inside ChatGPT?
No, it is a sample rather than a census. An advertiser missing from the library is not proof they are not running, it means we have not caught them yet. Coverage grows as the number of tracked prompts grows.
