ChatGPT ads cost: published pricing, CPM context from other channels, and first-test budgets

ChatGPT ads cost is controlled through bids and budgets you set, with billing tied to clicks or impressions and no public rate card. This guide walks a budget owner through how pricing works, what numbers OpenAI has published as of October 2026, how to frame CPM and CPC against your existing channels, and what a four-week pilot budget buys, before you design a full pilot in a separate test-planning workflow or implement tracking from a dedicated conversion-tracking guide.

What you pay for under each ChatGPT ads objective

ChatGPT ads offer three objectives, and each objective controls how you are billed. Per OpenAI's documentation as of October 2026, Views campaigns are billed per 1,000 impressions on a CPM (cost per mille) basis, Clicks campaigns are billed per valid click on a CPC (cost per click) basis, and Conversions campaigns optimize toward a supported conversion event while still billing either per valid click (oCPC) or per 1,000 impressions (oCPM).

You never pay per conversion, even when you choose the Conversions objective. The system uses conversion data to guide delivery but charges you only for clicks or impressions, depending on the optimization mode. This corrects a common misconception that conversion campaigns bill per conversion, which OpenAI's help content explicitly contradicts.

Bids are set at the ad-group level as a maximum CPC or maximum CPM, depending on the objective and optimization mode. The auction is a relevance-weighted second-price auction, so the price you pay is shaped by the next-best competing bid adjusted for relevance, not by your maximum bid itself. This means the misconception that your max bid is what you always pay is incorrect, because the system can clear at a lower price when competition is weaker.

Bid strength guidance in Ads Manager indicates whether your bid is likely competitive or may limit delivery, according to OpenAI's help content. As a budget holder, you can treat weak bid strength as a signal that the account team may need to raise bids or accept lower delivery at the current price ceiling.

Published bid guidance, budget floors, and billing mechanics

As of October 2026 OpenAI recommends starting with a maximum CPC bid of $3–5 for Clicks and Conversions campaigns. This is guidance, not a fixed price, but it gives you a reference range for first tests and for sanity-checking very low or very high bids proposed by your team. Since the auction is second-price, the actual CPC you pay can clear below your maximum when competition is lighter.

OpenAI enforces a minimum daily budget of $25 in USD, with documented equivalents such as 15 EUR, 15 GBP, 25 AUD, 25 CAD, 25 NZD, and 2,500 JPY. According to OpenAI's campaign creation documentation, daily budgets are treated as a 7-day delivery target, which means a single day can spend up to 2x the nominal daily budget, but the Sunday–Saturday week will not exceed 7x that daily amount. At $25 per day, the system can spend up to $50 on a high-delivery day, but will keep the total for the week at or below $175.

This 7-day mechanic also means a campaign that exhausts its weekly allowance can stop delivering while still marked active. As a budget owner, you should expect some day-to-day variability in spend and judge pacing on a weekly basis, not on any single day. OpenAI also notes that some budget-type changes are irreversible, such as switching from a campaign-total budget to a daily budget, and recommends starting with a daily budget, which aligns with the four-week pilot framing.

Billing itself follows the objective and optimization model: CPM for Views, CPC for Clicks, and oCPC or oCPM for Conversions, always tied to valid clicks or impressions. You never pay per conversion, so cost per acquisition is an outcome metric you compute from click or impression spend and tracked conversions, not a billing line item.

How to think about CPM and CPC context without a rate card

OpenAI does not publish a CPM rate card for Views campaigns, and per its documentation CPM pricing is discovered in the auction. There is therefore no official CPM figure you can budget against in advance, and any misconception that a published CPM exists should be corrected. Instead, you should treat CPM and CPC outcomes from ChatGPT ads as auction results that you compare to your own performance in other channels.

The reachable audience for ChatGPT ads is Free and Go plan users only, which OpenAI notes skews consumer. This audience shape should inform your expectations about both pricing and performance, because it differs from professional-only environments. TF practice is to avoid third-party CPM or CPC benchmark tables and to use your own trailing 90-day CPM and CPC from accounts you already run as the comparison point, since that is the number ChatGPT ads competes against for the same budget.

In practical terms, once you have a week or more of delivery, you can compute realized CPM and CPC from ChatGPT ads and line them up against your trailing 90-day averages in channels like Google, Meta, Microsoft, LinkedIn, Reddit, or Pinterest. This lets you judge whether ChatGPT is buying attention at a higher or lower unit cost than your existing mix, without relying on external benchmarks that do not match your audience or creative.

First-test budget math and what it buys

OpenAI's documented minimum daily budget of $25 sets the floor for any meaningful test. At that floor, a 30-day test spends about $750 in media (30 days multiplied by $25 per day). Using the recommended starting maximum CPC range of $3–5, that $750 budget buys roughly 150–250 clicks, calculated as $750 divided by $5 (150 clicks) and $750 divided by $3 (250 clicks).

If you are comfortable with a higher initial allocation, a $50 daily budget over the same 30 days spends about $1,500. At the same $3–5 maximum CPC guidance, that $1,500 buys roughly 300–500 clicks, computed as $1,500 divided by $5 (300 clicks) and $1,500 divided by $3 (500 clicks). These ranges come directly from arithmetic on OpenAI's published budget minimums and bid recommendations, and they define the scale of traffic you can expect from a first test.

TF practice is to treat ChatGPT ads as an early-stage channel, with budgets sized as pilots that clear the platform minimums and produce a readable four-week sample, but capped so a null result costs a defined test line item. The four-week test read is structured as follows: week 1 checks delivery (impressions and spend pacing against the daily budget, plus bid-strength guidance), week 2 checks click quality (click-through rate and landing-page engagement, including UTM-segmented behavior versus other paid channels), and weeks 3–4 check cost per conversion against existing channels' trailing 90-day numbers. TF practice is to compare weekly totals, not daily, because the 7-day budget mechanics and spend reporting lag distort daily reads.

If your unit economics require a certain number of conversions to judge success, you can work backward from these click volumes and your expected conversion rate. For example, if you expect a 5% conversion rate, 200 clicks from a $750 test would yield about 10 conversions, which is often enough for a directional comparison against your existing channels' cost per conversion.

Reading spend and performance with reporting lag

OpenAI's documentation states that impressions, clicks, and click-through rate refresh roughly every 15 minutes in Ads Manager. Spend insights, however, can lag by 7–8 hours, which means the spend number you see in the interface can be behind the actual charges that have accrued. This lag makes the misconception that day-one spend of zero means the test is not charging incorrect, because spend may simply not have surfaced in reporting yet.

Because of this lag, you should avoid making intra-day budget decisions based on apparent zero or low spend in the first hours of a campaign. TF practice is to compare weekly spend totals when reading performance, which aligns with both the 7-day budget mechanics and the several-hour delay in spend reporting. For conversion outcomes, OpenAI notes that conversion events require the pixel and/or Conversions API to be configured, and that UTM parameters persist on ad clicks so independent analytics can isolate ChatGPT traffic.

As a budget holder, you can ask your team to report weekly on three layers: delivery (impressions and spend versus the planned weekly cap), traffic quality (click-through rate and on-site behavior by UTM source), and cost per conversion relative to your trailing 90-day benchmarks in other channels. This cadence respects the reporting lag and the weekly budget structure, and it prevents overreacting to noisy daily fluctuations.

TF practice, approved in October 2026, is to size ChatGPT ads as four-week pilots that clear OpenAI's minimum budgets, use your own trailing 90-day CPM, CPC, and cost-per-conversion as the comparison set, and read results on a weekly cadence that matches the platform's 7-day budget mechanics and spend reporting lag.