ChatGPT Ads pricing: bids, budgets and what a campaign really costs

Three buying models: CPM, CPC and oCPC

OpenAI currently offers three campaign objectives, which set both the optimization and the billing:

Historically, the first phase of the pilot only allowed CPM buying; CPC was added afterwards so advertisers could align spend more directly with the actions people take after seeing an ad.

Where the bid is set: the ad group

The point most content misses: the maximum bid is set at the ad group level, not the campaign. Concretely, you enter a maximum CPM bid for a Reach campaign, or a maximum CPC bid for a Clicks campaign.

Ads Manager may also show bid-strength guidance indicating whether your bid is likely to be competitive or may limit delivery. That's the first signal to check when a campaign won't get off the ground, before concluding you have a targeting problem.

Budget, by contrast, is steered at campaign level: the two layers control different things.

Maximize Results: automatic bidding, now the default

On top of the manual bidding described above, OpenAI has added a second option: Maximize Results, a strategy that adjusts the bid itself to get as many results as possible (clicks or conversions, depending on the chosen objective) within the budget. It is now selected by default on new eligible ad groups: if you change nothing, it is the one steering delivery.

What it does not do, stated plainly by OpenAI: Maximize Results guarantees no target CPA, CPC or ROAS. It prioritizes the volume of results, not a precise cost per result. In practice, an account that wants to steer a precise cost per click or per conversion is better off switching back to manual bidding; an account that wants volume first, without micromanaging unit cost, can leave it active.

The setting is changed in the ad group's "Bid strategy" section, where manual CPC/CPM bidding remains available at any time. A detailed comparison, including with oCPC, is on our ChatGPT Ads Maximize Results page and our CPM vs CPC vs oCPC vs Maximize Results table.

OpenAI's recommended starting bid

This is the only official number on the cost of a click: for CPC campaigns, OpenAI recommends starting with a maximum bid of $3 to $5 USD per click.

Read that recommendation for what it is: a starting point for entering the auction, not a forecast of what you'll pay. On a second-price or dynamically-priced auction, the maximum bid caps your cost, it doesn't set it. And on young inventory, competition varies sharply by theme.

Conversely, a bid set too low is a mundane cause of complete non-delivery: without clearing the competitive threshold, the campaign simply wins no impressions. If you see zero impressions, check the bid-strength guidance before rewriting your context hints.

Budget: campaign total, daily budget and the applicable minimum

When creating a campaign, OpenAI offers two budget modes:

• A campaign-total budget: the total amount you want to spend over the full duration of the campaign. The documentation calls it the stricter control for total spend.
• A daily budget: an amount per day. Important detail, it is a delivery target, so actual spend may fluctuate above or below the selected amount on an individual day.

If your constraint is accounting ("don't exceed X this quarter"), the campaign-total budget is the right tool. If your constraint is a steady delivery pace, the daily budget fits better, as long as you accept day-to-day fluctuation.

OpenAI documents a minimum daily spend of $25 per campaign. That minimum varies by your ad account's billing currency, with a Minimum Campaign Spend table published in the Help Center.

Two practical consequences. First, a credible test cannot be run at 5 units of currency a day: the floor is structural, not negotiable. Second, since billing currency is permanently locked at account creation, so is the minimum that applies to your account. One more reason to take the ChatGPT Ads account setup screen seriously.

What changed: the $200,000 minimum is gone

A lot of content published in the first half of 2026 still cites a $200,000 minimum spend as the price of entry. That floor, in force at the February 2026 launch, was removed on May 5, 2026 with the opening of US self-service. Any approved advertiser can now set whatever test budget suits them, subject to the daily minimums above.

Economically, that is the program's most structural change: it moves ChatGPT Ads from a channel reserved for large accounts to one a small business can test. If a source still tells you today that you need $200,000 to get in, it has not been updated since May 2026.

How you actually get charged

Billing is not prepaid, it runs on a payment threshold: each account is assigned one, and the saved card is charged once unpaid spend reaches it. With a $25 threshold, the card is charged each time $25 of unpaid spend accrues.

OpenAI notes that approved advertisers typically start with a lower threshold, which may be increased automatically over time based on successful payment history and other factors. When you add the card, your bank may show a temporary $100 authorization during verification: that is not a charge.

The consequence to know: if a payment fails after the threshold is reached, delivery may stop until the issue is resolved. An expired card kills a campaign as effectively as an exhausted budget.

oCPC: optimizing for conversions without paying per conversion

oCPC is the most frequently misunderstood model. It optimizes delivery toward a conversion event you select, but billing does not change to pay per conversion: OpenAI charges you only when a valid click occurs. The CPA bid you enter is an optimization input, not a conversion charge.

The documented prerequisites are strict:

• The ad account must support conversion bidding.
• Conversion tracking must be set up, with the JavaScript pixel, the Conversions API, or both.
• You must have exactly one active standard conversion event to use as the optimization goal. Custom events cannot be oCPC optimization goals.
• You cannot change the goal or event after campaign creation, and an existing campaign cannot be converted to oCPC: you have to create a new one.

The technical setup is covered in our guide to the ChatGPT Ads pixel and conversion tracking.

Calculating your cost per lead or cost per acquisition

No platform can hand you your cost per lead: it depends on your post-click conversion rate, and so on your offer and your page. The formula is universal: CPL = CPC ÷ conversion rate.

The reading is instructive: at identical CPC, a 5x swing in conversion rate produces a 5x swing in CPL. Optimizing your landing page carries as much weight as negotiating your bid. That is exactly what our ChatGPT Ads landing page checker tests.

What OpenAI says about industry benchmarks: nothing reliable

Let's be direct, because this is where misinformation is densest: OpenAI itself notes there is no reliable, generalized performance benchmark by industry yet. The program is in beta and only a few months old in most markets.

The "average CPC by industry on ChatGPT Ads" tables you'll come across come from third-party market observations, often on the US pilot, with unpublished sample sizes. They can serve as an order of magnitude for building a hypothesis, never as the basis for a performance commitment to a client.

Our position: build your model from your own conversion data on other channels, test CPC within OpenAI's recommended range, then replace your assumptions with real numbers as soon as you have a few hundred clicks.

What reporting gives you to steer with

Ads Manager Beta reporting includes impressions, clicks, spend, click-through rate (CTR), average CPC, average CPM and conversions, available in table view, chart view and CSV export.

Two clarifications that keep your profitability math honest:

• The main Conversions column includes click-through conversions only.
• View-through conversions, shown for eligible accounts, use a fixed one-day window and are separate supplemental reporting. OpenAI states they should not be added to Conversions or used for core performance metrics like CPA.

Adding the two columns together is the mistake that makes a campaign look 30% more profitable than it is.

Building a realistic test budget

A simple method, resting only on documented reference points:

1. Start from the recommended bid, $3 to $5 per click, as your cost assumption.
2. Respect the daily minimum for your currency ($25 per campaign in USD).
3. Set a click volume large enough to conclude anything: at an expected 3% conversion rate, you need several hundred clicks before the number means something.
4. Choose a campaign-total budget if you want a hard ceiling.
5. Let it run: OpenAI recommends allowing at least 24 hours after launch before flagging a delivery issue, and reviewing enough volume before making large changes to bids, budgets or creative.

A test killed after six hours because "it isn't working" measures nothing except its own impatience.

What is not publicly documented

Points we flag as unconfirmed for lack of an official source:

• The actual average CPC or CPM, blended or by industry, is not published by OpenAI.
• The precise auction mechanism (first price, second price, other) is not publicly described, including for Maximize Results.
• The initial payment threshold amount and the pace at which it increases are not quantified.
• No guarantee of placement, volume or outcome is offered by OpenAI, and nobody can offer one on its behalf.

The full table of daily minimums by currency (dollar, pound, real, Mexican peso, yen, won...) is now gathered on a dedicated page: ChatGPT Ads minimum budget by country.