The two pricing models explained
ChatGPT ads now offer two distinct ways to pay: CPM and CPC. Understanding the mechanics of each is the foundation for every bidding decision you’ll make on the platform.
CPM: pay per thousand impressions
CPM (cost per mille) was the original pricing model when OpenAI launched ChatGPT ads in February 2026. You pay a fixed rate for every 1,000 times your ad is shown, regardless of whether anyone clicks. Current CPM rates range from $25 to $60, with most advertisers landing between $35 and $50 depending on topic competitiveness and relevance scoring.
CPM is inherently a reach-first model. You pay for exposure, not engagement. This makes it ideal for brand awareness campaigns, product launches, and situations where you want maximum visibility within specific conversation topics. The risk is straightforward: if nobody clicks, you still pay full price.
$25–$60
CPM range on ChatGPT ads as of May 2026
CPC: pay per click
CPC (cost per click) was added to the ChatGPT ads platform in April 2026, in response to performance advertisers who wanted to pay only for measurable engagement. Starting bids range from $3 to $5, though competitive topic clusters can push bids higher. You pay only when a user clicks your ad, which means zero-click impressions cost you nothing.
CPC is a performance-first model. You tie spend directly to user action, which makes budget forecasting more predictable and ROI calculation more straightforward. The trade-off is that you may pay more per impression than you would under CPM, especially if your ads generate high click-through rates.
$3–$5
Starting CPC bid range on ChatGPT ads
Side-by-side comparison
| Dimension | CPM | CPC |
|---|---|---|
| You pay for | Impressions (per 1,000) | Clicks only |
| Typical cost | $25–$60 per 1K impressions | $3–$5 per click |
| Available since | February 2026 | April 2026 |
| Best for | Reach, awareness, high-CTR campaigns | Direct response, lead gen, cost control |
| Risk profile | Pay even if no clicks | May overpay when CTR is high |
| Budget predictability | Predictable spend, variable outcomes | Variable spend, predictable cost per action |
Why OpenAI added CPC
The addition of CPC was a deliberate move to attract performance-focused advertisers. CPM-only pricing created a barrier for direct-response marketers and smaller businesses who needed to tie every dollar to a measurable outcome. By adding CPC, OpenAI opened the platform to advertisers who would never have tested ChatGPT ads under a CPM-only model.
The strategic calculus is clear: OpenAI wants both brand advertisers (who prefer CPM for reach) and performance advertisers (who prefer CPC for accountability). Offering both models lets them capture the full spectrum of advertising demand, similar to how Google offers both CPC search ads and CPM display ads.
When CPM wins
CPM is not inherently better or worse than CPC. It wins in specific scenarios where the math favors paying for impressions over paying for clicks.
High CTR scenarios
The fundamental logic is simple: if your click-through rate is high enough, CPM gives you cheaper clicks than CPC. The break-even point depends on your CPM rate and your CPC bid, but the general rule is that once your CTR exceeds roughly 1%, CPM starts to outperform CPC on a cost-per-click basis.
Here’s the math. At a $42 CPM and a 1.5% CTR, you get 15 clicks per 1,000 impressions. Your effective cost per click is $42 ÷ 15 = $2.80. That’s cheaper than the $3–$5 CPC bid you’d pay under the click-based model. The higher your CTR climbs, the wider this gap becomes.
$2.80 effective CPC
At $42 CPM with 1.5% CTR – cheaper than $3–$5 CPC bids
Best use cases for CPM
Brand awareness campaigns. When your primary goal is getting your brand in front of decision-makers during relevant conversations, you want maximum impressions. CPM guarantees delivery volume regardless of click behavior.
Testing new context hints. When you’re experimenting with new topic clusters and don’t yet know which ones will generate clicks, CPM lets you gather impression-level data without the risk of spending nothing (and learning nothing) on clusters that happen to have low initial CTR.
Competitive topics where you already perform well. If you’ve established that your ad creative resonates with users in certain conversation types and your CTR consistently exceeds 1%, switching those campaigns to CPM locks in a lower cost per click.
Break-even CTR calculator
The table below shows the effective CPC at different CPM rates and CTR levels. Find your CPM rate on the left, then scan across to see at which CTR the effective CPC drops below your CPC bid.
| CPM Rate | 0.5% CTR | 1.0% CTR | 1.5% CTR | 2.0% CTR | 2.5% CTR |
|---|---|---|---|---|---|
| $25 | $5.00 | $2.50 | $1.67 | $1.25 | $1.00 |
| $35 | $7.00 | $3.50 | $2.33 | $1.75 | $1.40 |
| $42 | $8.40 | $4.20 | $2.80 | $2.10 | $1.68 |
| $50 | $10.00 | $5.00 | $3.33 | $2.50 | $2.00 |
| $60 | $12.00 | $6.00 | $4.00 | $3.00 | $2.40 |
Reading the table: if your CPM is $42 and your CPC bid would be $4, CPM becomes cheaper once your CTR exceeds roughly 1.05%. At 1.5% CTR, you’re saving $1.20 per click. At 2.0% CTR, you’re saving $1.90 per click. Those savings compound quickly at scale.
When CPC wins
CPC is the safer choice in most scenarios, especially for advertisers who are new to ChatGPT ads or running campaigns in topic clusters where they don’t yet have CTR baselines.
Direct response campaigns
When your goal is driving clicks to a landing page, sign-up form, or product page, CPC aligns your spend with your objective. You pay only when someone takes the action you care about (clicking through), which means zero budget is spent on impressions that don’t generate engagement.
This is particularly valuable for lead generation, free trial sign-ups, demo requests, and e-commerce product pages where every click has a calculable value. If you know your average conversion rate from click to purchase is 3% and your average order value is $150, you can work backwards to determine your maximum acceptable CPC.
Low-CTR topic clusters
Some conversation topics generate high impression volume but low click-through rates. This is common in early-stage educational conversations where users are exploring a category but aren’t ready to engage with a vendor. Under CPM, these impressions eat your budget without delivering clicks. Under CPC, they cost you nothing.
$8.40 vs $4.00
Effective CPC at 0.5% CTR with $42 CPM vs a $4 CPC bid – CPC saves 52%
The math is stark. At 0.5% CTR and a $42 CPM, your effective cost per click is $8.40. Under a $4 CPC bid, you’d pay exactly $4 for the same click – a 52% savings. For campaigns with sub-1% CTR, CPC is almost always the better choice.
First campaigns without CTR baselines
If you’re launching your first ChatGPT ad campaign and have no historical CTR data, CPC eliminates the risk of overpaying for impressions. You can gather two weeks of click and impression data under CPC, calculate your actual CTR, and then use the break-even table above to determine whether CPM would be cheaper. This data-first approach prevents the common mistake of choosing CPM based on optimistic CTR assumptions that don’t materialize.
Cost comparison at different CTR levels
| CTR | CPM Cost (at $42) | Effective CPC | CPC Bid ($4) | Winner |
|---|---|---|---|---|
| 0.3% | $42 for 3 clicks | $14.00 | $12.00 | CPC |
| 0.5% | $42 for 5 clicks | $8.40 | $20.00 | CPC |
| 0.8% | $42 for 8 clicks | $5.25 | $32.00 | CPC |
| 1.0% | $42 for 10 clicks | $4.20 | $40.00 | ~Even |
| 1.5% | $42 for 15 clicks | $2.80 | $60.00 | CPM |
| 2.0% | $42 for 20 clicks | $2.10 | $80.00 | CPM |
The break-even point at a $42 CPM and a $4 CPC bid is approximately 1.05% CTR. Below that, CPC wins. Above that, CPM wins. This is the single most important number for your bidding decision.
The auction: how bids actually work
Understanding the auction mechanics is essential because your bid amount is not the only factor that determines whether your ad gets shown or what you actually pay. ChatGPT uses a relevance-weighted, second-price auction that differs from both Google’s and Meta’s systems in important ways.
Relevance-weighted scoring
When a user’s conversation matches multiple advertisers’ topic clusters, ChatGPT doesn’t simply award the impression to the highest bidder. Instead, it calculates an ad score that combines your bid with a relevance score. The relevance score is determined by how well your ad’s context hints, creative copy, and landing page align with the specific conversation happening right now.
Relevance is weighted more heavily than raw bid amount. This means a $3 CPC bid with high relevance can beat a $5 CPC bid with moderate relevance. OpenAI has stated that this weighting is intentional – they want ads to feel like useful additions to the conversation, not intrusions. The practical implication: improving your ad quality and context hint precision is often more cost-effective than raising your bid.
1 ad per response
ChatGPT shows a single ad per conversation turn – no competing sidebar or footer ads
Second-price auction mechanics
ChatGPT uses a second-price auction, meaning you pay just enough to beat the next-highest-scoring competitor, not your maximum bid. If your ad score would have won at $3.20 and your bid was $5, you pay roughly $3.20 (adjusted for relevance weighting). This is similar to how Google’s ad auction works and encourages advertisers to bid their true willingness to pay rather than gaming the system.
One ad per response
Unlike Google (which shows 3–4 ads per search results page) or Meta (which intersperses multiple ads in a feed session), ChatGPT displays exactly one ad per conversation response. This has two implications for bidding: first, winning the auction gives you exclusive attention with no competing ads in view. Second, the auction is winner-take-all for each impression – there is no second or third position to fall back to.
How context hint quality affects cost
Context hints are the topic-level signals you provide to tell the auction which conversations your ad is relevant to. Well-crafted context hints increase your relevance score, which means you can win auctions at lower bids. Broad, generic context hints lower your relevance score, forcing you to compensate with higher bids.
For example, a CRM advertiser with the context hint “CRM software for small service businesses with under 20 employees” will score higher in matching conversations than one with “business software.” The specific advertiser may win the auction at $3.50 while the generic advertiser needs to bid $5.50 for the same impression.
ChatGPT vs Google Ad Rank
| Factor | Google Ad Rank | ChatGPT Ad Score |
|---|---|---|
| Bid weight | Moderate – bid × Quality Score | Lower – relevance weighted more heavily |
| Quality signal | Quality Score (CTR, relevance, landing page) | Relevance score (context hints, creative, landing page) |
| Positions available | 3–4 per results page | 1 per response |
| Pricing type | Second-price (generalized) | Second-price (relevance-adjusted) |
| Primary optimization lever | Keyword selection and bid management | Context hint precision and creative quality |
Bidding strategy by goal
The right bidding model depends on what you’re trying to achieve. Here’s a decision matrix that maps campaign goals to recommended models, starting bids, and key performance indicators.
| Goal | Model | Starting Bid | Primary KPI |
|---|---|---|---|
| Lead generation | CPC | $4–$5 | Cost per lead |
| Free trial sign-ups | CPC | $3–$4 | Cost per sign-up |
| Brand awareness | CPM | $30–$45 | Branded search lift |
| Product launch | CPM | $35–$50 | Impression volume, direct traffic |
| E-commerce sales | CPC | $3–$5 | ROAS |
| Competitive conquest | CPM | $40–$60 | Share of voice, CTR |
| Demo requests (B2B) | CPC | $4–$5 | Cost per qualified demo |
| Content distribution | CPC | $3–$4 | Cost per page view |
Budget allocation: the 70/30 split
Regardless of which bidding model you choose, allocate your ChatGPT ad budget in a 70/30 split: 70% to proven strategies and topic clusters, 30% to experimentation. The proven portion sustains your baseline performance. The experimental portion tests new context hints, creative angles, and bidding models that could become your next proven winners.
This split prevents the common trap of either being too conservative (missing opportunities) or too aggressive (burning budget on unproven tactics). Review the split monthly and graduate winning experiments into the proven bucket.
When to switch models
Switch from CPC to CPM when your CTR exceeds the break-even threshold for two consecutive weeks. The consistency matters – a single high-CTR day doesn’t justify the switch. Switch from CPM to CPC when your CTR drops below the break-even point for two consecutive weeks, or when you’re entering a new topic cluster where you lack CTR data.
Never switch both the bidding model and the creative simultaneously. If you change two variables at once, you cannot attribute performance changes to either one. Change the bidding model, observe for two weeks, then iterate on creative if needed.
Cross-platform pricing comparison
Advertisers who run campaigns across multiple platforms need to understand how ChatGPT pricing compares to Google, Meta, and LinkedIn. The raw numbers are only part of the story – conversion rates and audience quality change the cost-per-conversion calculus significantly.
| Platform | Avg CPC | Avg CPM | Avg CVR | Effective CPA |
|---|---|---|---|---|
| ChatGPT (CPC) | $3–$5 | N/A (click-based) | 2.0–4.0% | $100–$250 |
| ChatGPT (CPM) | ~$4.20 (at 1% CTR) | $25–$60 | 2.0–4.0% | $105–$210 |
| Google Ads | $2–$5 | $100–$1,000 | 1.0–2.0% | $100–$500 |
| Meta Ads | $0.50–$3 | $5–$20 | 1.0–1.5% | $33–$300 |
| LinkedIn Ads | $5–$12 | $30–$80 | 1.5–2.5% | $200–$800 |
Why raw CPC comparison is misleading
Looking at CPC alone, ChatGPT appears comparable to Google and more expensive than Meta. But CPC ignores the most important variable: what happens after the click. Early advertiser data shows ChatGPT ads generating 1.5–4× higher conversion rates than Google Ads in matching verticals. Users who click a ChatGPT ad have already self-qualified through their conversation – they’ve described their problem, specified their requirements, and are actively evaluating solutions.
This means the cost-per-conversion comparison tells a very different story than the cost-per-click comparison. A $4 CPC with a 3% conversion rate produces a $133 cost per conversion. A $3 Google CPC with a 1.5% conversion rate produces a $200 cost per conversion. The “cheaper” click is actually the more expensive conversion.
1.5–4× higher CVR
ChatGPT ad conversion rates vs Google Ads in matching verticals
Lapis advantage: one prompt, four platforms
The practical challenge of advertising across ChatGPT, Google, Meta, and LinkedIn is creative production. Each platform has different format requirements, character limits, and creative best practices. Lapis generates ad creatives for all four platforms from a single product description, so you can test cross-platform campaigns without the production bottleneck. For a detailed breakdown of how ChatGPT compares to Google and Meta on targeting, measurement, and user behavior, read our ChatGPT vs Google vs Meta ads comparison.
How to optimize your bids over time
Bidding optimization is not a set-it-and-forget-it decision. The right approach evolves as you accumulate data. Here’s a phased framework for the first two months.
Week 1–2: establish baselines with CPC
Start every new campaign with CPC at $4 per click. This gives you a predictable cost structure while you gather the data you need: CTR by topic cluster, click quality (bounce rate, time on page), and conversion rate from click to desired action. Don’t optimize during this phase. The goal is data collection, not performance maximization.
Run at least 3–5 ad variations to give the relevance engine enough signal to optimize delivery. Use Lapis to generate multiple headline and description variations from a single brief so you’re not bottlenecked by copywriting capacity.
Week 3–4: evaluate CPM opportunity
After two weeks, calculate your actual CTR across each topic cluster. For any cluster where CTR exceeds the break-even threshold (use the calculator table above), run a parallel CPM test. Keep the CPC campaign running as a control – allocate 70% of the cluster’s budget to CPC and 30% to CPM. Compare effective CPC across both models after 1,000+ impressions on each.
Month 2: A/B test bidding models at scale
With four weeks of data, you can make informed decisions about which clusters should run on CPC vs CPM. Split your budget across models based on where each performs better. The clusters with CTR above the break-even point run on CPM. Everything else stays on CPC.
During this phase, start optimizing the creative quality lever. Better ads produce higher relevance scores, which reduce your cost-per-impression (under CPM) or cost-per-click (under CPC) through the auction mechanics described above. Small creative improvements compound: a 10% CTR improvement on a CPM campaign drops your effective CPC by 10%.
Lapis forecasting for bid optimization
Lapis performance forecasting can model the expected CTR for different creative variations before you spend budget running them. This shortens the optimization cycle: instead of running 10 variations for two weeks to find winners, you can use forecasting to narrow the field to 3–4 high-probability winners and allocate your test budget more efficiently. Combined with the break-even calculator, this lets you predict which bidding model will perform better for each creative variation before launch.
For a comprehensive optimization framework covering creative iteration, targeting refinement, and performance benchmarks, see our ChatGPT ads optimization playbook.
CPA bidding: what’s coming
OpenAI has announced that cost-per-acquisition (CPA) bidding is in development and expected to launch later in 2026. CPA will let advertisers set a target cost per conversion and let the platform’s algorithm optimize delivery to hit that target. This is the same model that powers Google’s Target CPA and Meta’s Cost Cap strategies.
Why CPA matters
CPA removes the need to manually calculate break-even points between CPC and CPM. You tell the platform what a conversion is worth to you, and the algorithm decides whether to optimize for clicks or impressions on a per-auction basis. For performance advertisers, this is the end state – the platform handles the bidding math so you can focus on creative quality and conversion rate optimization on your landing page.
How to prepare now
Set up conversion tracking today. CPA bidding requires a feedback loop between your website and the ad platform. The sooner you implement conversion tracking, the more historical data the CPA algorithm will have to work with when it launches. Read our ChatGPT ads conversion tracking guide for step-by-step setup instructions.
Define your target CPA now. Calculate your current cost per conversion under both CPC and CPM models. This gives you a realistic baseline for setting CPA targets when the feature launches. Advertisers who set CPA targets based on historical data consistently outperform those who guess.
Build a conversion history. Run campaigns now under CPC to accumulate conversion data. The more conversion events the algorithm has to learn from, the faster CPA bidding will optimize when it becomes available. A minimum of 50 conversions per campaign is generally needed for algorithmic bidding to stabilize.
The evolution from CPM-only to CPM plus CPC to CPM plus CPC plus CPA mirrors the trajectory of every major ad platform. Google started with CPC, added CPM for display, then introduced CPA (Target CPA) and eventually automated bidding (Maximize Conversions). Meta followed a similar path. ChatGPT ads are moving through the same stages, just faster. Advertisers who understand the bidding fundamentals now will have a structural advantage as the platform matures.
Ready to test CPC vs CPM on your next ChatGPT ad campaign? Start with Lapis to generate high-relevance ad creatives that lower your auction costs, then use the break-even calculator above to pick the right bidding model for each topic cluster. For a full walkthrough of campaign setup, see our complete guide to ChatGPT ads.
