ChatGPT Ads in 2026

ChatGPT Ads went live on 9 February 2026, opened to self-serve on 5 May, and passed a $1 billion annualised run rate in under 200 days. Self-serve buying reached India on 31 August. Here is what they cost, how the targeting actually works, and the limitations worth knowing before you commit budget.

ChatGPT Ads went live on 9 February 2026, opened to self-serve advertisers on 5 May, and reached a $1 billion annualised revenue run rate in under 200 days. Self-serve buying opened in India, Europe, the Middle East and North Africa on 31 August 2026. There is no minimum spend.

That is the fastest any major ad platform has gone from launch to a billion-dollar run rate. This covers what the ads actually look like, who sees them, what they cost, how targeting works, and whether they are worth your budget — with the limitations stated as plainly as the opportunities.

What a ChatGPT ad actually is

There is currently one format: the chat card. It is a sponsored card that appears below ChatGPT's answer, containing a title, a short description, an image and a destination link. It sits in a subtly tinted box, clearly labelled, visually separated from the response above it.

The critical design decision is that ads do not influence the answer. OpenAI has been explicit that the model's response is generated independently, and the sponsored card is appended afterwards based on the topic of the conversation. Whether users perceive that separation is a different question, but architecturally the answer is not for sale.

If that sounds familiar, it should. A titled card with an image and a link, matched to surrounding content rather than to a keyword, appearing below editorial material — that is structurally a native ad. The interface is new; the format is not.

Who sees ChatGPT ads

This is the single most important eligibility fact and the one most guides bury.

Ads reach logged-in adult users on the Free and Go tiers only. Everyone paying $20 a month or more never sees an advertisement.

How the rollout happened

What ChatGPT ads cost

Two things stand out. The CPM more than halved in three months as inventory opened up, which is the normal trajectory for a new platform moving from scarcity pricing to auction pricing. And the CPC floors are high — $3 to $5 depending on category.

Putting those CPCs in context

ChatGPT ads are priced like search, not like native or social. That is defensible — conversational intent genuinely resembles search intent more than feed browsing — but it sets a high bar. At a $4 CPC and a 4% landing page conversion rate, your cost per lead is $100. That only works if a lead is worth considerably more than that.

How targeting works

This is where ChatGPT ads differ most from everything that came before, and it is genuinely new rather than a rebrand of existing targeting.

Placement is matched to conversational context — the topic of the current conversation, past chat history and previous ad interactions — rather than to keywords you bid on or to cross-web behavioural profiles. You are not buying "personal loan" as a keyword. You are buying relevance to conversations about borrowing money.

The auction is relevance-weighted, so final cost depends on competition, creative quality and landing page relevance rather than a fixed rate. If that sounds like Google's Quality Score, it works similarly in effect — better creative and a more relevant destination lower your effective cost.

The measurement limitation nobody warns you about

OpenAI shares no individual conversation data with advertisers. Reporting is aggregated by design — described as closer to connected TV measurement than to Google Ads in its level of detail.

For a performance marketer used to Google Ads, this is a significant adjustment. You will not see the query that triggered your ad. You will not get granular conversation-level attribution. You get aggregate delivery and performance data, and you infer the rest.

These are architectural privacy decisions rather than early-stage gaps, so waiting for them to be fixed is not a strategy. Plan around aggregate measurement and lean harder on your own analytics and CRM to close the loop.

ChatGPT ads in India

Self-serve buying opened across India on 31 August 2026. A few things matter for Indian advertisers specifically.

The account country designation is generally permanent once set. Getting it wrong at signup is not a quick fix, so decide deliberately whether you are an Indian advertiser targeting India or an Indian agency buying for overseas clients before creating the account.

The addressable audience is Free and Go tier users. In a price-sensitive market, that skews the available audience toward the larger, less affluent segment — which is good for reach and less good if your offer needs high purchasing power.

The CPC floors are dollar-denominated. A $3 to $5 floor is roughly ₹265 to ₹440 per click. Against Indian native inventory at ₹1 to ₹4 a click, that is two orders of magnitude more expensive. For most Indian lead-generation offers, the arithmetic simply does not work yet.

The honest conclusion for most Indian advertisers: ChatGPT ads make sense today if you sell something expensive. For high-volume, low-value lead generation, native inventory at ₹1 to ₹4 a click remains dramatically more efficient.

How to set up a campaign

1. Create an advertiser account at the OpenAI Ads Manager. Set the country designation carefully — it is generally permanent.
2. Choose your bidding model. CPM suits awareness and reach objectives; CPC suits direct response. Most performance advertisers should start on CPC to cap downside.
3. Write the card. Title, short description, image, destination. The constraints are tight, so specificity beats cleverness.
4. Match the landing page to the conversation. The auction is relevance-weighted, and landing page relevance is an explicit factor in what you pay.
5. Set aggregate success criteria before launch, because granular attribution is not available.
6. Budget for a learning period. A new platform with a new targeting model needs volume before its performance means anything — the same principle that applies everywhere else.

Creative that works in this format

The chat card appears after the user has already received an answer. That is a fundamentally different moment from a search result, where the ad appears before the answer. Your card is not competing to answer the question — it is offering the next step after the question was answered.

Practically: creative that reads as "here is the thing you now need" outperforms creative that reads as "here is the answer". The user already has the answer. Sell the action.

ChatGPT ads vs everything else

Is it worth it

Test it if your conversion is worth several hundred dollars or more, your buyer is likely on a free or low-cost tier, you can tolerate aggregate-only reporting, and you have budget that is genuinely experimental rather than load-bearing.

Wait if you run high-volume low-value lead generation, your reporting depends on granular attribution, your buyer is a senior professional likely paying for Plus or Enterprise, or your unit economics need clicks under a dollar.

The organic side: answer engine optimisation

Running ads inside ChatGPT is only half the picture. The other half is whether ChatGPT mentions you in the answer itself, which no amount of ad spend buys.

Answer engine optimisation — sometimes called AEO or GEO — is the practice of structuring content so AI systems cite it. The mechanics overlap heavily with good SEO: clear structure, factual specificity, schema markup, and genuine authority on a topic. The difference is that being cited in a synthesised answer matters more than ranking at a position.

Brands investing in AEO are better positioned when advertising opens more broadly, for a simple reason. A sponsored card sitting beneath an answer that already mentions you converts considerably better than one appearing beside an answer that recommends a competitor.

What happens next

The direction of travel is clear enough from the first 200 days. One format will become several. Geographic expansion is continuing market by market. A $1 billion run rate in under 200 days means the commercial incentive to expand inventory is enormous, and inventory expansion usually means ads appearing in more places within the product.

The open question is user tolerance. OpenAI has been careful about labelling and answer independence precisely because the format only works while people trust the answers. If that trust erodes, so does the inventory's value. Every advertiser buying this placement is making a bet on OpenAI holding that line.

ChatGPT Ads is a new platform changing quickly — eligibility, pricing and features have already changed several times since launch. Details reflect publicly available information as of September 2026. Verify current terms in the OpenAI Ads Manager before planning budget.