How much do ChatGPT ads cost in 2026?

What’s the average ChatGPT ads CPM?

Observed CPMs sit around $25 as of June 2026, based on the published reports listed in the sources below. OpenAI’s launch default was $60, which is what you charge when you have no idea what your inventory is worth and would like the market to tell you. The market told them: reported clearing prices fell to roughly $25 within 4 months.

Two caveats on that number. It is an average across context categories, and the platform is young enough that month-to-month movement of 20% or more is normal. Commercial-intent contexts, where someone is comparing products, cost more than ambient conversation, so your blended CPM depends on which conversations you win. Treat $25 as a planning input, and replace it with your own clearing prices after your first 2 weeks of data.

Should you bid CPC, CPM, or for conversions?

OpenAI’s Ads Manager supports all three: CPM from the February launch, CPC since May, and conversion-optimized CPA bidding in early access since June 5. Most performance advertisers should start with CPC at $3 to $5. Paying for clicks caps your downside while CPMs are volatile, and it hands the platform a clean signal about which impressions earn attention.

CPM starts to win once your click-through rate is proven. At a $25 CPM, a 1% CTR works out to $2.50 per click and a 2% CTR to $1.25, both under a $3 CPC. The break-even point is a CTR of about 0.8%. Below that, CPC is cheaper and you should stay on it. Above it, CPM lets you buy the impressions directly and stop paying the platform’s per-click margin.

Conversion bidding is the third step, not the first. It needs a conversion event already flowing through the pixel or CAPI, so it pays off after you have tracking and a couple of weeks of data, not on day one. When it works, it changes what you are buying: instead of bidding a price for a click and hoping it converts, you tell the platform your target cost per acquisition and let it find the conversions. The setup guide covers the order to do this in, and the catch is real: garbage conversion data produces a confidently wrong bid.

What does a realistic test budget look like?

Here is the math on $2,500 per month at a $25 CPM. That buys 100,000 impressions. At a 1% click-through rate you get 1,000 clicks, an effective $2.50 per click. If 2% of those visitors convert, that is 20 conversions at a $125 cost per acquisition.

Swap in your own assumptions and the structure holds: budget over CPM gives impressions, CTR turns impressions into clicks, and conversion rate turns clicks into a CPA you can judge. The arithmetic is deliberately simple because the inputs are the uncertain part. A CTR of 0.5% instead of 1% doubles your CPA to $250 on the same budget, which is why the first two weeks are about learning your real CTR before you scale the spend.

What CPA can you expect for your business?

The honest answer is that the CPA is the same arithmetic for everyone. Whether it is good news is entirely about your economics. The $125 CPA above is a disaster for one business and a bargain for another. Three worked cases from the same $2,500 budget and 20 conversions make the point.

• Ecommerce, $40 average order: 20 orders is $800 in revenue against $2,500 spent. A $125 CPA on a $40 order loses money on every sale. This advertiser needs a CTR or conversion rate roughly 4 times higher, or a much larger average order, before the channel pays.
• SaaS, $50 per month subscription: 20 trials at a $125 CPA is viable if those subscriptions last. At $50 a month, a customer covers the $125 in under 3 months, so the test works as long as churn is reasonable and the 20 are real subscriptions, not free trials that lapse.
• B2B, $5,000 contract: 20 leads at a $125 cost per lead is strong before you even discuss close rate. Close 1 in 10 and you have spent $2,500 to win $10,000 in contracts. For this advertiser the right question is why the test budget is so small.

The pattern: ChatGPT’s current prices favor businesses with high order values or strong retention, the same as every other paid channel. Model your own version of this table before you launch, not after.

How do you forecast spend before you launch?

Work backward from the result you need, not forward from a budget you picked. Start with a target cost per acquisition your margins can carry, then use the same chain in reverse: how many conversions do you want, at what conversion rate, at what click-through rate, at a $25 CPM. The arithmetic tells you the budget, and it usually tells you something uncomfortable.

Say you want 50 conversions a month at a $100 target CPA. That is $5,000 in spend if you hit the target. At a 2% conversion rate, 50 conversions needs 2,500 clicks; at a 1% click-through rate, 2,500 clicks needs 250,000 impressions; at a $25 CPM, that is $6,250. The two numbers disagree because your assumed CPA and your assumed rates imply different worlds, and the gap is exactly the thing your first two weeks exist to measure. Forecast with ranges, fund the test, and replace the assumptions with your own data before you scale.

What drives ChatGPT ad costs up or down?

Context category moves the price first. Ads are matched to conversation topics rather than keywords, and commercial-intent contexts (someone asking ChatGPT to compare project-management tools) draw more bidders than ambient conversation does. You have some control here through which contexts your creative is written to win.

Creative fit moves your effective cost second. The auction weighs your bid against how well your ad is likely to fit, so a more relevant ad can clear a slot at a lower price. An ad that reads like a useful continuation of the conversation earns clicks and lowers your real cost. An ad that reads like an interruption gets scrolled past, and weak engagement makes every impression you bought work less hard.

Competition is the variable to watch through 2026. Self-serve opened in May and the bidder pool is still thin, so current clearing prices are a tenant’s-market rate. The same scarcity that keeps prices low also makes early data noisy, which is the trade you accept for getting in before the auction fills.

What will ChatGPT ads cost as the auction fills in?

More than they do now, in the direction if not the exact number. The fall from $60 to $25 was a young auction with few bidders clearing low, not a structural discount, so the honest projection is that CPMs rise as more advertisers arrive through 2026 and into 2027. We do not have a dated forecast worth publishing, and we will not invent one.

What this means for planning is straightforward. The cheap inventory is a window, not a baseline, so the value of testing now is partly the prices and partly the head start: an advertiser who learns the context model and accumulates conversion data while the auction is thin holds priced-in creative when it is not. We update this page when the numbers move, and the changelog at the top tracks each change.

How do ChatGPT ad costs compare to your other channels?

Run the comparison against your own accounts rather than industry averages. If your search campaigns pay $4 to $6 per click, ChatGPT’s $2.50 effective clicks (at a $25 CPM and 1% CTR) price in cheaper, with intent quality somewhere between search and social. If your paid-social CPMs run under $15, ChatGPT impressions cost more, and the case rests on context: your ad appears inside an answer someone asked for.

The fuller head-to-heads (auction depth, intent quality, measurement maturity) are their own pieces: ChatGPT ads vs Google ads and ChatGPT ads vs Meta ads. The complete guide covers the rest of the channel. The short version: the channel prices like an early one, because it is one, and the numbers on this page will drift as that changes.