What Are ChatGPT Ads? The Complete 2026 Guide for Advertisers
A plain-English guide to ChatGPT ads in 2026: the ad format, who sees ads, how context hints target them, typical costs, policies and who should test them first.
ChatGPT ads are sponsored placements that appear inside ChatGPT conversations, directly below the assistant's answer. They look like a compact card with a brand name, a short headline, a line of description, an image and a link. They are labelled as sponsored and they sit visually apart from the answer itself.
That simple description hides a big shift. For the first time, a brand can show up at the exact moment someone asks an AI assistant for help with a problem the brand solves. This guide covers how the format works, who actually sees it, how targeting differs from search and social, what it costs, and how to decide whether it belongs in your 2026 media plan.
A short history: how ChatGPT ads got here
OpenAI began showing ads to logged-in adult users on the Free and Go tiers in the US on February 9, 2026. In the early months, access was reportedly limited to large advertisers, with minimum commitments widely reported around $200,000.
That changed on May 5, 2026, when OpenAI opened a self-serve ChatGPT Ads Manager (in beta) to US advertisers. There is no minimum total spend, and the minimum daily budget is about $25 per campaign in the US, though it varies by market. Since then OpenAI has added CPC bidding, bulk edits, CPM-to-CPC conversion, product-feed and carousel formats, and an optimized CPC (oCPC) option in beta for product-feed campaigns. Ads have expanded beyond the US and were reported live in 40-plus countries by late 2026, including Brazil and Mexico, both announced in August.
For a dated, step-by-step record of each change, see our ChatGPT ads timeline.
What a ChatGPT ad looks like
OpenAI calls the unit a chat card. Every card has the same building blocks:
| Element | What it is | Limit (as of mid-2026) |
|---|---|---|
| Advertiser name and favicon | Pulled from your account and site | Fixed |
| Title (headline) | The main line people read | 3 to 50 characters |
| Body (description) | One supporting sentence | About 100 characters |
| Image | A single visual | Square-friendly image |
| Landing URL | Where the click goes | Must match the ad |
The card shows up under ChatGPT's response, clearly marked as sponsored. OpenAI has stated that ads do not change ChatGPT's answers. The answer is the answer; the ad is a separate, optional next step.
That separation matters for how you write. You are not interrupting a feed or competing with ten blue links. You are offering one relevant option to someone who just asked a specific question. If you want to check your copy against these limits before you launch, the ad character counter handles it in seconds.
Who sees ChatGPT ads
Only logged-in adults on the Free and Go plans see ads. Go is a low-cost tier, around $8 a month, available in more than 170 countries. Paid tiers (Plus, Pro, Business, Enterprise and Edu) are ad-free.
That single rule shapes who ChatGPT ads are good for:
- Good fit: consumer brands, DTC and ecommerce stores, consumer apps, course creators, creators and prosumers, and local or consumer services. Their buyers are well represented among Free and Go users.
- Poor fit: B2B companies selling to teams. Many professional buyers use ChatGPT through Business or Enterprise seats, or pay for Plus, so they never see an ad.
ChatGPT is reported at roughly 900 million weekly users, so even the ad-supported slice is a large audience. We break down the tier details in who sees ChatGPT ads.
How targeting works: context hints, not keywords
This is the part most advertisers get wrong at first. ChatGPT ads are not targeted with keyword lists or demographic checkboxes. They are targeted with context hints: plain-language descriptions of the conversations and situations your ad belongs in.
A weak hint reads like a keyword: "running shoes." A strong hint describes a person, a situation and a need: "new runner training for a first 10K who gets shin pain and wants more cushioned shoes." ChatGPT matches that description against real conversations. It may also use what it remembers across a person's chats (memory, where enabled) to pick relevant ads.
Good hints usually combine four things:
- Situation: what is happening in the person's life right now.
- Need: the problem they want solved.
- Use case: how they will use the product.
- Intent: how close they are to buying.
If you want a starting point, the free context hints generator turns a product description into situation-based hints you can edit. Our full context hints guide goes deeper on testing and refinement.
What ChatGPT ads cost
You pay OpenAI directly for media. Pricing options, as of mid-2026:
- CPM: you pay per thousand impressions. This was the original model. Third-party observers have reported CPMs roughly in the $25 to $45 range, which is high compared with social feeds.
- CPC: you pay per click. Added with self-serve. OpenAI's recommended starting bids work out to roughly $3 to $5 per click.
- oCPC: optimized CPC, in beta for product-feed campaigns from August 2026, where the system adjusts bids toward conversions.
Here is a hypothetical example. Say you set a $50 daily budget on CPC and pay an average of $4 per click. That buys about 12 clicks a day. If your landing page converts at 3%, you would expect roughly one sale every three days at that budget. If your average order is $90, that is about $30 of revenue per day on $50 of spend, a 0.6 ROAS on first purchase. The point of the math is not to scare you off. It shows that budget, click cost, conversion rate and order value all have to line up, and that repeat purchases often decide whether a channel works.
Run your own numbers with the ChatGPT ads budget calculator. For a full breakdown of bids and budgets, read how much ChatGPT ads cost.
Why the intent is different
Search ads catch people who type a few words. Social ads catch people who are scrolling. ChatGPT ads catch people mid-thought, often after they have described their situation in detail.
Someone might write: "I have a 9-month-old who keeps waking at 4 a.m., we've tried an earlier bedtime, what else can we try?" That is far richer context than a search for "baby sleep." An ad for a blackout sleep sack that answers that specific problem is not an interruption. It is a relevant option.
Early third-party data supports the idea that this traffic is valuable. Criteo reported in February 2026 that traffic from ChatGPT converted about 1.5 times better than other channels. Similarweb reported an average CTR around 0.68%, with top brands near 1.57%. Treat these as directional, not guarantees for your account.
The rules: what you cannot run
OpenAI's ad policies are stricter than many advertisers expect. Not allowed:
- Deceptive or unverifiable claims, invented statistics and fake reviews
- Guaranteed outcomes, such as health cures or income promises
- Before-and-after transformations
- Copy that exploits hardship or insecurity, or implies the ad knows what the user said
- Impersonating OpenAI, ChatGPT, public figures or other brands, or using other companies' trademarks
- Prohibited products, including drugs, weapons, tobacco and vapes, counterfeits, adult content, gambling, hacking tools and payday loans
- Clickbait, ALL CAPS and fake urgency
Health and supplements, wellness, financial products, alcohol, dating, weight management and products for minors are restricted, which means extra review and narrower wording. Political ads are out of scope. And your landing page must match what the ad promises.
Before you submit, the ChatGPT ad policy checker flags common problems. Our full ChatGPT ad policies guide covers each category.
How measurement works
OpenAI provides a pixel (often called the OAIQ pixel) that tracks browser events such as page views, add to cart, sign-ups, leads and purchases. There is also a server-side Conversions API, and automatic advanced matching rolled out in August 2026. Third-party analytics integrations, such as Triple Whale and Hightouch, have been reported.
Set up both the pixel and the Conversions API from day one. Without conversion data, CPC bidding is guesswork, and you will not be able to compare ChatGPT against your other channels.
Should you test ChatGPT ads now?
ChatGPT ads are worth testing if most of these are true:
- You sell to consumers, creators or prosumers, not teams.
- Your buyers ask AI assistants for advice before they buy.
- You can afford a test of roughly $50 to $200 a day for several weeks.
- Your product and landing page can make honest, verifiable claims.
- You already have messages that sell on other channels.
That last point is underrated. The fastest way to a working ChatGPT ad is not a blank page. It is a message you already know converts, rewritten as a direct answer to a buyer's question. The longest-running ads in your niche on Meta are a strong signal of which promises, proof and offers keep selling.
How SecondWin helps
SecondWin is a done-for-you ChatGPT ads service. You enter your URL and we read your site to work out your product, buyer and niche. We then study the longest-running ads in your niche in Meta's Ad Library, pull out the proven promise, proof and offer, and map them to the questions your buyers ask ChatGPT. From there we write original chat cards and context hints, check each one against OpenAI's policies and our own rules, and launch and manage campaigns inside your own OpenAI ad account. Ad spend is billed by OpenAI to you directly.
Start with the free URL analysis to see the buyer questions and ad angles for your brand, then compare plans and pricing. SecondWin is independent and not affiliated with OpenAI.
FAQ
Are ChatGPT ads the same as sponsored answers?
No. The ad is a separate chat card that appears below ChatGPT's response and is clearly labelled as sponsored. OpenAI has stated that ads do not change what ChatGPT says in its answer. Advertisers cannot pay to be mentioned inside the response itself. The card is an optional next step for the user, shown because its context hints match the conversation, and it links to the advertiser's own landing page.
Do Plus and Pro users see ChatGPT ads?
No. As of 2026, ads are shown only to logged-in adult users on the Free and Go tiers. Plus, Pro, Business, Enterprise and Edu plans are ad-free. This is why ChatGPT ads tend to suit consumer brands far better than B2B companies selling to teams, since many professional buyers use ChatGPT through paid or workplace plans.
What is the minimum budget for ChatGPT ads?
Since self-serve Ads Manager opened in May 2026, there is no minimum total spend. The minimum daily budget is about $25 per campaign in the US and varies by market. Most advertisers will need more than the minimum to gather useful data quickly, which is why a starting range of $50 to $200 a day per campaign is a common recommendation.
Can I target ChatGPT ads by keyword?
Not in the traditional sense. ChatGPT ads use context hints, which are plain-language descriptions of the situations and conversations your ad belongs in. A phrase like "dinner recipes" works poorly. A hint like "parent looking for a quick weeknight dinner their kids will eat" works much better, because it describes a person, a need and an intent that ChatGPT can match.
Are ChatGPT ads available outside the US?
Yes. Ads launched in the US in February 2026 and have since expanded. They were reported live in more than 40 countries by late 2026, with Brazil and Mexico among the markets announced in August. Minimum daily budgets and available features can vary by country, so check Ads Manager for your market before planning a launch.