ChatGPT ads: current formats, availability, and account requirements

Ads in ChatGPT are a real, live channel as of October 2026, but with narrower availability and formats than mature platforms like Google or Meta. This guide focuses on whether you can use it, what the ad units look like, and which constraints matter before you invest time, with deeper setup detail covered in a separate setup walkthrough.

What ads in ChatGPT are and how they appear

Per OpenAI’s documentation, ChatGPT ads are bought through Ads Manager (Beta) at ads.openai.com and appear directly inside conversations as sponsored content. The current live placement is a labeled ad shown below the assistant’s response, not a scripted conversational chatbot experience.

The ad unit is a static card that contains the advertiser name, a favicon or logo, a title (headline), body copy (description), a landing page link, and an image asset. This corrects a common misconception that ChatGPT ads are interactive chatbot flows you can script; in reality, the unit is a conventional card format rendered under the model’s answer.

OpenAI applies placement safeguards that exclude sensitive user contexts, including mental or personal health and emotionally reliant conversations, and brand-unsafe contexts such as political content, regulated goods, and violence. This means you are buying in-conversation attention, but not in every possible conversation, and some of the most sensitive or polarizing topics are explicitly off-limits for ad serving.

Which users see ChatGPT ads and in what situations

As of October 2026, ads are shown only to ChatGPT users on Free and Go plans. Users on Plus, Pro, Business, Enterprise, and Edu plans do not see ads at all, and accounts identified as under 18 are excluded from ad delivery based on account age information and age prediction.

This directly corrects the misconception that everyone who uses ChatGPT sees ads. The audience is structurally consumer-skewed, because paid professional tiers and under-18 users are removed from the ad-eligible pool by design.

Within eligible accounts, ads appear in-conversation below responses, subject to the contextual safeguards described earlier. You do not choose specific conversations; instead, OpenAI’s systems decide when a given exchange is suitable for an ad impression.

What information a ChatGPT ad unit contains

The standard ChatGPT ad card has 6 components, all of which go through policy review before serving:

Per OpenAI’s ad policies, every one of these elements, plus the landing page itself, is reviewed before an ad is allowed to run. Ads or landing pages that OpenAI’s systems cannot review are not eligible, which matters if your site uses formats or gates that interfere with automated review.

This structure also corrects the misconception that a ChatGPT ad is a way to inject your product directly into organic answers. The ad card is clearly separated, labeled, and clickable, and feed-based products are explicitly limited to ad placements during beta rather than organic responses.

Where advertisers must be based to use Ads Manager

OpenAI’s Ads Manager self-service access is limited to businesses in 63 listed countries as of October 2026. The list includes the US, UK, Canada, Australia, most of the EU, Japan, Korea, Brazil, Mexico, India, Singapore, the UAE, Saudi Arabia, Turkey, and Vietnam, among others.

The legal entity that advertises and is billed must be based in one of these supported countries. This corrects the misconception that any business can sign up for ChatGPT ads as easily as for Google or Meta; entity location is a hard requirement, not just a targeting choice.

In newly opened European markets, advertisers can also access ChatGPT ads through OpenAI’s Ads Solutions team, agency partners, or technology partners. For some new self-serve accounts, campaigns may initially be limited to the advertiser’s home country, with eligibility to target other countries unlocking only after identity verification approval and a required spend threshold in home-country campaigns.

Which categories can advertise, and which cannot

During the initial period, OpenAI’s ad policies state that ads are primarily limited to consumer verticals. These include lifestyle and household goods, local services, travel and experiences, and digital products and education.

Financial services, healthcare and medicine, and legal services are treated as regulated categories. Ads in these verticals may be approved on a case-by-case basis, US-only, with manual review, and OpenAI may require proof of licensure before allowing campaigns to run.

Several categories are disallowed at launch. These include adult content and dating, alcohol and tobacco, gambling, political content, recreational drugs including cannabis, counterfeit goods, scams, individual housing or job listings, and unsubstantiated wellness claims. OpenAI’s policy changelog shows active movement from version 1.0 in March 2026 through version 1.6 in September 2026, including legal services opening up in August 2026, so category eligibility is evolving over time.

This policy framework further corrects the misconception that any business can sign up and advertise freely. Many sensitive or high-risk categories are either banned or tightly constrained, especially outside US consumer contexts.

What an advertiser account must have to get started

Advertisers sign up at ads.openai.com and go through an advertiser review process that checks business legitimacy, including verification, account quality, and risk signals. After that, every ad’s title, copy, media, and landing page are reviewed before serving, and creatives that cannot be reviewed are not eligible.

Per OpenAI’s campaign creation help, there is a minimum daily budget of 25 USD, with currency-specific floors for 24 currencies such as 15 EUR, 15 GBP, and 2,500 JPY. This minimum applies at the campaign level and defines the smallest daily spend you can allocate to test the platform.

Some new self-serve accounts, especially in newly opened markets, may initially be allowed to advertise only in their home country. Eligibility to expand targeting to other countries is contingent on identity verification approval and achieving a required spend threshold from campaigns that target the home country.

Current formats and what is in beta

The live format available to all eligible advertisers is the standard ad card described earlier, with name, logo, headline, description, image, and landing page link. This is the format that appears below ChatGPT responses for Free and Go users.

In beta, OpenAI offers product feed campaigns for retail. Advertisers upload a catalog via CSV, hosted URL, or SFTP, and product ads can show images, titles, star ratings, prices, sale prices, and brand information. Per OpenAI’s documentation, feed products are eligible for ads only during beta and explicitly do not appear in organic answers, which corrects the misconception that product feeds can push items into non-sponsored ChatGPT responses.

Hotel feeds also exist as a limited beta in the developer documentation. These feeds are a separate format from standard cards and retail product feeds, and are currently documented only as a constrained test.

How ChatGPT ads are priced at a high level

OpenAI describes three campaign objectives: Views, Clicks, and Conversions. Views use CPM (cost per 1,000 impressions), Clicks use a valid-click CPC (cost per click), and Conversions use either oCPC (optimized cost per click) or oCPM (optimized cost per 1,000 impressions).

You never pay per conversion; spend is still on impressions or clicks even when you choose a conversion objective. OpenAI recommends a starting maximum CPC bid of 3 to 5 USD, and auctions use a relevance-weighted second-price model, which means you pay based on the next-best bid adjusted for relevance rather than your full bid amount.

For deeper coverage of pricing mechanics, bid strategies, and first-test budgets, see the dedicated cost and bidding guide.

Key limitations compared with Google and Meta

The first limitation a Google or Meta advertiser will notice is targeting. There is no keyword bidding in ChatGPT ads; instead, targeting is based on conversational context plus ad-group “context hints,” which OpenAI defines as guidance signals rather than exact-match keywords. Per OpenAI’s documentation, context hints do not guarantee placement in any conversation, which corrects the misconception that they work like traditional keyword bidding.

The second limitation is audience composition. Because ads are shown only to Free and Go users, and explicitly not to Plus, Pro, Business, Enterprise, Edu, or under-18 accounts, the reachable audience is intentionally consumer-skewed and excludes many paid professional users.

Reporting is also described as relatively thin compared with mature platforms. Available metrics include impressions, clicks, spend, click-through rate, average CPC, average CPM, and conversions, with metrics refreshing roughly every 15 minutes but spend figures lagging by 7 to 8 hours.

There are structural constraints on campaign configuration. You cannot change a campaign’s objective after creation, and some budget-type changes are irreversible once set. Custom audience inclusion requires at least 25,000 matched users, and OpenAI advises that campaigns targeting the European Economic Area and Switzerland avoid custom audiences while personalization is unavailable there.

For a broader view of what the platform can and cannot do relative to other channels, see the ChatGPT ads capabilities overview, and for measurement setup details refer to the conversion tracking guide.