There isn't a single number that answers "what do ChatGPT ads cost," because OpenAI has already changed its own pricing model once since launch. Ads went live in February 2026 at roughly 60 CPM. As that price eroded to as low as $25 in some categories, OpenAI shifted toward a cost-per-click model, and by the time the self-serve Ads Manager opened in May, advertisers were bidding directly, typically in the $3 - $5 per click range. If you're building a budget today, plan around the bidding model that's actually live, not the launch pricing that made headlines in February.
This guide breaks down what a realistic ChatGPT ad budget looks like in practice, what it buys you at different spend levels, and how it stacks up against the Google and Meta budgets you're probably already running.
The Short Answer
| ChatGPT Ads | Google Search Ads | Meta (Facebook) Ads | |
|---|---|---|---|
| Pricing model | CPC (shifted from CPM) | CPC | CPC |
| Typical cost | ~$3–$5 per click | ~$2–$5.42 per click (varies heavily by industry) | ~$0.60–$0.70 (traffic), ~$1.80–$1.92 (lead gen) |
| Minimum daily budget | $25/day (no account minimum since self-serve launch) | No platform minimum; realistic minimums vary by industry | No platform minimum; realistic minimums vary by industry |
| Audience reached | Free and $8/mo Go tier users only | Anyone searching on Google | Anyone on Meta platforms |
| Maturity | Under a year old, pricing still being tuned | 20+ years of auction data | 15+ years of auction data |
The headline: ChatGPT ads currently cost more per click than Meta and are roughly in line with, or slightly above, a mid-range Google Search click but you're bidding into a much newer, less competitive auction, which cuts both ways. Less competition can mean more efficient reach today. It also means today's price and rules are the least stable of the three platforms.
How the Pricing Actually Works Right Now
Campaign budgets in the self-serve ChatGPT Ads Manager can start as low as $25/day, and the $200,000 account minimum that gated the original invite-only pilot is gone. That makes the platform technically accessible to almost any advertiser. But accessible isn't the same as useful: a $25/day budget at $3–$5/click buys somewhere between 5 and 8 clicks a day, which isn't enough volume to learn anything statistically meaningful about what's working. A more realistic starting point for an actual test is $100–$300 per day, sustained for about two weeks, which gives you enough clicks to compare performance across a couple of ad variations and context-hint configurations before deciding whether to scale.
Bidding itself works the way most CPC auctions do: you set a maximum bid per click, OpenAI's system matches your ad against relevant, ad-eligible conversations based on topic, general location and language, and you're only charged when someone clicks. There's no exact published formula for how ad rank is calculated beyond bid amount, but early advertiser reporting suggests relevance to the conversation and creative quality both factor in, similar to how Google's Quality Score has worked for years.
What a Realistic Starting Budget Looks Like by Business Size
Small business or solo test ($750–$1,500/month): Enough to run the recommended $100/day test window for two weeks, then a further two weeks at a reduced rate once you know which creative and targeting combination performs. This is a "find out if this channel works for us" budget, not a growth budget.
Mid-market ($3,000–$8,000/month): Enough to run $200–$300/day sustained, which supports testing multiple context-hint variations (different audience framing, different product angles) in parallel and gathering enough click volume per variant to compare conversion rates, not just CPC.
Enterprise or aggressive pilot ($10,000+/month): At this level you're less budget-constrained and more constrained by the addressable audience remember that Plus, Pro, Business, Enterprise and Education subscribers never see these ads at all, so spend beyond what the Free/Go tier audience in your target categories can absorb is wasted regardless of budget size.
That last point is worth repeating because it's the single most common budgeting mistake brands will make with this channel: sizing a ChatGPT ad budget the way you'd size a Google Ads budget, without checking first whether your actual buyers are even in the ad-eligible tiers.
The Categories That Can't Buy In at Any Budget
Before you build a budget at all, check eligibility. OpenAI's ad policies currently exclude several categories outright: dating, alcohol, tobacco, healthcare, financial services, legal services, gambling, and political content. Eligible categories include household and consumer goods, local services, travel and experiences, and digital products and education. If your business sits in an excluded category, the budget conversation is moot no amount of spend buys you into this channel right now, which means your entire AI visibility strategy for ChatGPT has to run through organic citations. (See Vista's AI citation tracking if that's your situation it's worth knowing how you already show up before assuming you have no options.)
What Targeting Actually Costs You in Practice
ChatGPT Ads Manager targeting works across three layers: platform (Android app, Android web, iOS app, iOS web, desktop web), location (country-level globally, with U.S. campaigns able to narrow to state, DMA or ZIP code), and context hints short natural-language constraints like "for a team under 20 people" or "budget under $10,000" that narrow which conversations your ad is eligible to appear in. The more specific your context hints, the smaller your eligible audience and, generally, the higher your effective cost per useful click, because you're fishing in a smaller pool. Broad targeting keeps your nominal CPC lower but risks matching conversations only loosely related to your offer. Budget for a narrowing pass: start closer to broad, watch which context-hint variations actually convert, then tighten.
How This Compares to Google and Meta, Beyond the Raw CPC
Raw CPC comparisons are useful but incomplete. Google Search ads cost more per click on average than Meta because they capture people already searching with commercial intent you're paying for intent, not just attention. Meta is cheaper per click because it's interrupting a social feed rather than answering a direct question. ChatGPT ads sit in an interesting middle position: they appear after someone has already asked a specific, often high-intent question, which argues for Google-like value, but the platform is brand new, so advertiser competition and therefore price discipline in the auction hasn't matured the way it has on Google after two decades. Early advertisers are effectively paying "beta platform" prices for what could turn out to be "high-intent search" quality traffic. That's the actual bet you're making with a ChatGPT ads budget right now, and it's worth sizing your test spend as exactly that: a bet, not a proven channel.
Hidden Costs Beyond the Bid
The CPC or CPM number is only part of the real cost. Budget time and resources for writing and testing ad creative and context hints specifically for a conversational format, which reads differently from a search ad headline or a social feed post. Budget for proper attribution setup — without distinct UTM parameters and conversion tracking for ChatGPT traffic, you won't be able to tell whether a 3–5 click is actually worth it against your other channels. And budget, mentally if not financially, for the opportunity cost of treating ad spend as a substitute for the organic work that determines whether you're recommended inside an answer at all — a topic covered in more depth in ChatGPT Ads Are Here: What They Mean for Your Brand's AI Visibility Strategy.
How Costs Vary by Category
The 3–5 blended CPC range is a starting reference point, not a number you should expect in your specific category. Travel and experiences, one of the eligible categories, tends to carry higher purchase values and correspondingly higher advertiser willingness to bid, which likely pushes real CPCs above the blended average during peak booking seasons. Digital products and education, by contrast, often see lower average order values and more price-sensitive advertisers, which tends to pull category CPCs down. Local services sit in between, with costs that vary enormously by geography the same way local Google Ads costs do a home services company bidding in a major metro will see different economics than the same business in a smaller market.
The practical implication: don't budget off a single industry-wide number you read in an article, including this one. Run your own two-week test at the 100–300/day level, look at what your actual cost-per-click and cost-per-conversion come out to inside your category, and use that as your real benchmark going forward. Blended averages are useful for setting expectations before you spend anything; they're not a substitute for your own data once you have it.
A Simple Framework for Setting Your Budget
Start by establishing your current organic citation baseline with a tool like Vista's AI Rank Tracker before spending a dollar on ads you need a "before" number to know whether ad spend is adding incremental reach or just paying for exposure you'd have gotten for free.
Then run a genuine two-week test at $100–$300 per day rather than a token $25/day budget that won't generate learnable data. Compare your blended cost per lead from that test against your existing Google and Meta numbers using the same attribution window, not just the raw CPC.
If ChatGPT ads clear your existing cost-per-lead bar, scale gradually and keep testing context hints as the platform's targeting options mature. If they don't clear it, redirect that budget toward the AEO work that keeps paying off long after any given campaign ends. Vista's plans start at $99/month for teams that want to track that organic side properly.
