How much do ChatGPT ads cost? Budgets, bidding, and what a real test looks like

ChatGPT ads cost at least $25 a day, the minimum daily budget in OpenAI’s Ads Manager, and OpenAI suggests starting with a maximum cost per click around $3-5. What you actually pay depends on your category, how well your ad fits the conversations it gets matched to, and how good your conversion tracking is.

The minimum gets you into the auction. It won’t necessarily get you an answer. Here’s how pricing works as of fall 2026 and how to size a first test that does. (New to the channel? Start with what ChatGPT ads are and how they work.) Every example figure below is illustrative.

What is the minimum budget for ChatGPT ads?

The minimum daily budget is $25, which works out to $750 over a 30-day month. Treat it as the price of entry: at a few dollars a click, a campaign running at the floor buys only a handful of visits each day.

How does bidding work for ChatGPT ads?

You can bid three ways: CPM (pay per thousand impressions), CPC (pay per click), or oCPC, a conversion-optimized option that’s still billed per click. If you bid by click, OpenAI suggests starting with a maximum around $3-5.

• CPC is the natural starting point for lead and sales tests. You pay only for visits, and your max bid caps the price of a click.
• CPM pays for being seen. It suits brands that can measure lift some other way, and it’s harder to judge for a small business chasing leads.
• oCPC uses your conversion data to optimize toward conversions while you still pay per click. It’s in open beta, has to be enabled on your account, and only works as well as your tracking.

What drives what you actually pay?

Your real cost comes down to competition in your category, the relevance of your ad and landing page, and the quality of your conversion tracking. The minimum budget and suggested bid are the same for everyone.

Competition. You’re bidding against every other advertiser who wants the same conversations. If your category is expensive on Google, expect competition to follow you here as more advertisers arrive.

Relevance. ChatGPT matches ads using the conversation’s context and intent, your ad’s title, copy, and landing page, and the context hints you supply. A specific ad tied to a specific question gives the system more to work with than a generic one. Clicks from the wrong conversations are the most expensive kind, even when each one is cheap.

Tracking quality. OpenAI measures conversions with a JavaScript pixel, a server-side Conversions API, or both, deduplicated so a sale counts once. Running both is the sturdier setup, since a browser pixel on its own can miss conversions. If tracking is broken, you can’t calculate cost per lead, and oCPC has nothing reliable to optimize toward.

How big should a first ChatGPT ads test be?

Size the test backwards from the number of leads or sales you need to make a decision, then check that the daily budget clears the $25 minimum. For most businesses, that means well above the floor for more than a month.

In our guide to Google Ads budgets in Tampa, we use 15-20 leads as the working floor for a test you can actually read; below that, you can’t tell a good campaign from a lucky one. The same logic holds here.

What would a test look like for a Tampa Bay service business?

Say a hypothetical roofing company in Brandon sets a $40 daily budget for 45 days, targeting only the ZIP codes it serves in eastern Hillsborough County. Every number below is made up to show the arithmetic.

• Budget: $40 a day × 45 days = $1,800.
• Clicks: if clicks average $4, that’s $1,800 ÷ $4 = 450 clicks.
• Leads: if 1 in 25 visitors books an inspection, 450 ÷ 25 = 18 leads.
• Cost per lead: $1,800 ÷ 18 = $100.
• Cost per job: if the company closes 1 in 3 inspections, that’s 6 jobs, or $300 in ad spend per job.

Eighteen leads lands inside that 15-20 range, so there’s something real to judge. If a roof job is worth several thousand dollars, $300 per job looks healthy. If Google already produces comparable leads at, say, $60, ChatGPT has to win on lead quality or extra volume.

Now change one assumption. At $5 a click, the same $1,800 buys 360 clicks and about 14 leads, roughly $125 each. If only 1 in 50 visitors converts, $4 clicks produce 9 leads at $200 each. Small shifts in click price or conversion rate swing cost per lead hard, so let the test find your real numbers.

Location targeting keeps the test local. In the US, ChatGPT ads can target by state, metro area, or ZIP code, so the Brandon roofer never pays for clicks from Sarasota.

What should you measure?

Measure cost per lead or cost per sale, then compare it with what the same dollars buy on Google and Meta. Clicks and click-through rate tell you whether your ad is matched well; cost per customer tells you whether to keep spending.

• Cost per lead or sale, from ChatGPT’s conversion reporting, cross-checked against your CRM or store data.
• Lead quality. How many ChatGPT leads became customers, compared with Google and Meta leads. A cheap lead that never closes costs more than it looks.
• Attribution rules. ChatGPT ads use a 1-day view-through attribution window, and each platform counts conversions its own way, so dashboards won’t agree. Compare channels in one place under one set of rules. If your numbers already disagree, our analytics and reporting work starts there.

When should you not test ChatGPT ads yet?

Wait if you can’t track a lead or sale, if you can’t fund a test well past the minimum, or if your category is restricted. Until then, the money works harder elsewhere.

• No conversion tracking. Without the pixel or Conversions API installed and tested, you’ll buy clicks with no way to judge them.
• No budget for a real test. Two weeks at $25 a day costs $350 and is unlikely to produce enough leads to decide anything. Put that money toward your landing page or the channel that already works.
• A restricted category. Gambling and political ads aren’t allowed. Dating, alcohol, drugs, and health-claim ads are restricted, and finance, healthcare, and legal advertisers need case-by-case approval.
• Proven channels short on budget. If Google is profitable and capped only by spend, your next dollar probably belongs there.

We run ChatGPT ads tests with tracking set up before launch and results reported next to your other channels. If the shaky part is your Google, Meta, or Microsoft tracking, our free ads audit sends back the top 5 fixes within 3 business days of getting read-only access.

Common questions about ChatGPT ads costs

Is $25 a day enough for ChatGPT ads? It’s enough to run, since $25 is the minimum daily budget. At $3-5 a click, though, it buys roughly 5 to 8 clicks a day, so a test at the floor needs a long runway.

Are ChatGPT ads cheaper than Google Ads? There’s no reliable public benchmark yet, so your own test is the number that counts. Compare cost per lead or sale from ChatGPT with what Google produces for the same offer over the same period, under the same attribution rules.

How long should a ChatGPT ads test run? Long enough to collect the leads or sales you need to make a call, which for many service businesses means more than a month. Set the length from the math (budget, click price, conversion rate), and don’t judge it on the first week.