Can ChatGPT apps make money yet?

Why are ChatGPT apps suddenly worth taking seriously?

ChatGPT apps are worth taking seriously now because OpenAI has turned them from an experiment into a real distribution layer, even though the business model is still unfinished.

When OpenAI launched the Apps SDK, it said developers could build interactive applications that run directly inside ChatGPT and potentially reach more than 800 million weekly users. Early integrations included Canva, Spotify, Booking.com, Coursera, Expedia, Figma and Zillow. OpenAI then opened submissions to outside developers rather than keeping the ecosystem limited to launch partners.

The bigger change came later. Apps are now being folded into a broader Plugins system. OpenAI currently describes the Plugins Directory as the main place where users discover workflow capabilities across ChatGPT and Codex. A plugin can contain apps, reusable skills or both.

At the same time, OpenAI is preparing to retire custom GPTs. Its latest guidance tells GPT builders to migrate workflows toward Plugins, with the first Enterprise retirement already scheduled and other plans expected to follow.

That gives us a much clearer picture of where OpenAI is heading. The old idea of publishing a lightweight custom GPT and hoping it becomes popular is fading. The platform is moving toward software that connects ChatGPT to real services, data, transactions and workflows.

The opportunity has become more serious. It has also become harder.

Is a ChatGPT app basically the new GPT Store?

No. ChatGPT apps require much more substance than most custom GPTs did, and OpenAI is now actively moving away from the GPT Store model.

The GPT Store opened in early 2024 after users had already created more than three million custom GPTs. The barrier was extremely low. Builders could combine instructions, uploaded knowledge and actions without necessarily operating a substantial standalone product.

OpenAI initially said US builders would eventually earn money based on GPT usage. A limited experiment followed, but a broad creator-payment system never became the normal way GPT builders made money.

That history is especially relevant now because custom GPTs are being retired rather than becoming the foundation of OpenAI's next marketplace. OpenAI's current migration guidance points builders toward Plugins, which can combine instructions with actual connected apps.

This raises the barrier. A useful ChatGPT app today is more likely to have a backend, authentication, proprietary data, a useful workflow or access to a service that ChatGPT cannot reproduce by itself.

For serious developers, that may be good news. Three million easy-to-create GPTs produced plenty of supply but very little evidence of a healthy creator economy. A smaller ecosystem of apps that actually do things has a better chance of producing businesses people will pay for.

Earlier GPT Store model	Where OpenAI is heading now
Custom GPTs	Plugins containing apps and skills
Mostly instructions and configuration	Real services, APIs and workflows
GPT Store discovery	Plugins Directory plus conversational discovery
Usage payments discussed but never broadly deployed	No general usage revenue share today
Very low barrier to entry	More engineering and product work

Does OpenAI pay ChatGPT app developers when people use their apps?

No. There is still no broad system that pays ChatGPT app developers according to usage.

This is one of the easiest parts of the market to misunderstand. A developer can build an app, get it approved and have people use it without receiving any automatic share of their ChatGPT subscription.

OpenAI has talked about developer monetization several times. The GPT Store originally came with the promise of usage-based builder earnings. When Apps SDK launched, OpenAI again said monetization details would follow.

Yet OpenAI's current Apps SDK documentation still says those details will be shared in the future.

That means 100,000 app interactions do not currently translate into a predictable payout from OpenAI. There is no public standard rate per invocation, active user or minute of use.

This becomes painful for AI-native apps with real operating costs. A developer may be paying for databases, APIs, external models, storage or infrastructure every time someone uses the app while receiving nothing from OpenAI for that activity.

So a free ChatGPT app built around the idea that “we'll grow usage first and OpenAI will pay us later” still has no dependable business model behind it.

Can you charge people for a ChatGPT app now?

Yes, developers have more room to charge users now, although ChatGPT still lacks the simple native subscription system people usually imagine when they hear “app marketplace.”

This changed meaningfully during 2026. When public app submissions first opened, OpenAI's commercial rules were much more restrictive. Early transaction support focused mainly on physical goods, while digital subscriptions and other paid features were still expected later.

OpenAI's current App Developer Terms are broader. They explicitly allow a developer to send users to an external website controlled by the developer to complete a payment, as long as the transaction follows OpenAI's policies and applicable law.

That means a ChatGPT app can now sit in front of a normal software business with its own billing system. A user could discover the service through ChatGPT, use part of it there and then complete payment on the developer's own site.

OpenAI also says Apps SDK developers can connect their existing backends so customers can sign in and access premium features.

What we still do not have is a standard button that lets an unknown developer publish a $10-a-month ChatGPT utility and have OpenAI handle subscription billing, entitlements, renewals and revenue sharing like Apple does.

Revenue method	Current situation
OpenAI pays per app use	No broad program
Existing subscribers log into the app	Available
Developer sends users to external checkout	Allowed under current terms
Commerce checkout for supported products	Available
Native subscription sold by an independent ChatGPT app	No standard system
Advertising inside ordinary ChatGPT apps	No mature general model

Are businesses already making money through ChatGPT apps?

Yes. ChatGPT already generates real commercial transactions, with shopping providing the clearest proof.

OpenAI launched Instant Checkout with Etsy sellers and built the Agentic Commerce Protocol with Stripe. It later integrated Instacart deeply enough that a user could discuss a meal, build a grocery basket and complete checkout through ChatGPT.

The money in those transactions is real. OpenAI says merchants pay a small fee on completed Instant Checkout purchases, while merchants keep responsibility for payments, fulfilment, returns and the customer relationship.

Shopping inside ChatGPT has expanded since that first launch. OpenAI has also been building richer product discovery, comparison and merchant information into conversations, which makes ChatGPT increasingly useful earlier in the purchase journey rather than only at checkout.

This is proof that ChatGPT can create revenue. The limitation is simple: Etsy sellers, retailers and Instacart already had businesses before their ChatGPT integrations existed.

Their ChatGPT presence becomes another sales channel.

That tells us a lot about where the platform works commercially today: companies that already know what they sell and how they get paid can plug ChatGPT into that machine much more easily than a developer trying to create an entire business inside the chat.

Can an existing SaaS company make money from a ChatGPT app?

Yes. Existing SaaS companies currently have one of the clearest ways to make ChatGPT apps financially useful.

Imagine a software company already charging $30 per user each month. It does not need OpenAI to collect that $30 again. Customers can authenticate through the ChatGPT app and use features covered by the subscription they already have.

The ChatGPT integration only needs to do one of a few commercially useful things: bring in new customers, make current customers use the product more often, reduce churn or make the software easier to use.

Canva is a good example of the broader idea. Someone can move from discussing a presentation to creating something with Canva without treating ChatGPT and Canva as completely separate workflows. Figma brings design work into the same environment. Similar logic applies to project-management software, CRMs, analytics platforms and document tools.

The economics get even better in expensive B2B software. If a company pays $50,000 a year for a product, nobody needs to monetize each individual ChatGPT interaction. Making the existing contract more useful is enough.

That is why the strongest business case currently looks much closer to “ChatGPT as another interface for our product” than “our entire product exists only as a ChatGPT app.”

Does ChatGPT's huge user base mean your app will get traffic?

No. ChatGPT may have hundreds of millions of users, but an independent app can still receive almost no traffic.

This is probably the biggest trap in the whole opportunity.

When OpenAI launched Apps SDK, it highlighted the possibility of reaching more than 800 million ChatGPT users. Developers naturally notice that number. Yet OpenAI's current developer terms explicitly say it does not guarantee placement, visibility, ranking or promotion in the directory or inside ChatGPT conversations.

Early developer reports show how wide the gap can be.

In one OpenAI Developer Community discussion, an approved developer reported receiving between zero and five tool calls a day. Another developer said an app that was searchable but absent from the curated directory was getting roughly three to ten registrations a day, with much of that traffic apparently automated.

OpenAI support later confirmed that approval and prominent placement are separate things. Most newly approved apps do not automatically receive featured distribution. Apps that show useful real-world usage and good user satisfaction may eventually receive more exposure, but developers cannot simply apply for it.

Those community reports are anecdotes, so we should not turn them into an “average ChatGPT app traffic” statistic. But they expose a real structural problem.

A platform can have 800 million users and still give your app 10 users.

Can ChatGPT recommend your app automatically when someone needs it?

Yes, ChatGPT can surface relevant apps during conversations, but developers still cannot count on that discovery mechanism for reliable customer acquisition.

This is the feature that could eventually make ChatGPT dramatically different from a traditional app store.

Normally, a user needs to realize they need an application, search for one, compare options, install it and then learn how to use it. ChatGPT already knows what the person is trying to accomplish from the conversation. In theory, it can bring in the right app exactly when the user needs it.

OpenAI has been experimenting with this kind of contextual recommendation since Apps SDK launched. It has said that conversational context, app usage and user preferences can help determine which capabilities appear.

But developers have repeatedly asked how surfacing actually works. OpenAI support has explained that stronger real-world usage and user satisfaction can lead to broader distribution, including proactive recommendations. There is no public formula telling developers how much usage they need or how individual ranking factors are weighted.

The cold-start problem is obvious. An app may need strong usage before ChatGPT exposes it more widely, while getting strong usage is difficult when ChatGPT rarely exposes it.

For now, conversational recommendation is an acquisition channel worth testing. Building a revenue forecast around it would be reckless.

Is ChatGPT app discovery good enough to replace marketing?

No. A new ChatGPT app still needs its own way of attracting users.

The early app directory already showed why. OpenAI developers complained that searchable apps could remain almost invisible unless users knew what to look for. OpenAI support confirmed that only a limited set received stronger placement.

Search itself has also been relatively narrow. Earlier this year, OpenAI said app search intentionally relied heavily on names and subtitles because searching full descriptions had produced noisy results. That makes discovery harder for a developer whose brand is unknown.

The product has since moved toward the broader Plugins Directory, which should give OpenAI more ways to organize apps, skills and complete workflows. That may improve discovery over time. We do not yet have enough evidence to say the problem is solved.

OpenAI also supports direct links to app experiences. In practice, developers can promote those through websites, newsletters, social media, communities or their existing products.

So the economics are very different for two developers publishing identical apps. One already has 100,000 email subscribers. The other has no audience and expects ChatGPT to find all the users.

Today, we would give the first developer much better odds.

Which ChatGPT apps have the easiest path to revenue right now?

ChatGPT apps closest to an existing transaction have the easiest path to making money today.

Travel makes the distinction obvious. A generic travel-planning assistant may answer thousands of questions without generating a cent. An app that helps someone choose and book accommodation can attach itself to a transaction that already has economic value.

Shopping works the same way. A conversation about running shoes can end with an order. Grocery planning can end with an Instacart basket. A software recommendation can end with a subscription. A business workflow can push a company toward a larger SaaS contract.

At the other end are tools such as calculators, writing helpers, small image utilities, quizzes and novelty apps. They can be genuinely useful while having no natural payment moment.

Those products need another mechanism, usually an existing account, a premium subscription outside ChatGPT or enough ongoing value that users are willing to leave the conversation and pay.

The closer an app gets to money that users were already preparing to spend, the easier its economics become.

ChatGPT app category	Distance from a transaction	Monetization today
Shopping	Very close	Strong
Travel booking	Close	Strong where transactions can be captured
Existing SaaS	Close	Strong
B2B workflow software	Close	Strong with external contracts
Lead generation	Medium	Viable if leads convert
Standalone AI utility	Farther away	Harder
Entertainment or novelty	Far	Weak without large reach

Can a new AI app make money without sending users outside ChatGPT?

This remains the hardest ChatGPT app business model today because the product can work perfectly inside the conversation while the payment layer still lives somewhere else.

Imagine we build an excellent specialist app that analyses a document and returns exactly what the user needs. There is no reason for that person to visit another website. Keeping everything inside ChatGPT actually makes the product better.

Now look at the economics.

OpenAI does not broadly pay the developer for usage. There is still no standard native subscription system for third-party ChatGPT software. The developer may nevertheless be paying for servers, databases, licensed data or other APIs every time the app runs.

External checkout helps, but it introduces friction. We are asking the user to leave a smooth conversational workflow to create an account or pay elsewhere.

Existing SaaS companies barely notice this problem because their users already have accounts and subscriptions. A brand-new product whose entire experience belongs inside ChatGPT feels it immediately.

This is probably the clearest gap between ChatGPT's technical potential and its current developer economics.

Could a free ChatGPT app make money by generating leads?

Yes. Lead generation is one of the more realistic ways for an independent ChatGPT app to make money without charging users directly.

Consider mortgages, property, accounting, recruitment, business software, solar installation or home renovation. A user may never pay the ChatGPT app. Revenue can arrive later when that person requests a quote, books a consultation, signs up with a provider or completes a purchase.

ChatGPT is particularly interesting here because conversations can reveal detailed intent.

Someone searching Google for “solar panels” tells us relatively little. Someone telling a ChatGPT app, “I own a 160-square-metre house near Lyon, my electricity costs about €2,600 a year and I want solar next spring,” is giving a service provider a much clearer picture of whether that lead is useful.

Privacy rules obviously constrain what developers can collect and how they can use it. OpenAI requires apps to request information that is necessary for the service and to explain their privacy practices.

Within those limits, the economics can still work very well. A home renovation lead might be worth far more than a $10 monthly app subscription.

For expensive services, a relatively small amount of qualified ChatGPT traffic may therefore be enough to build a meaningful business.

Are B2B ChatGPT apps easier to monetize than consumer apps?

Often, yes. B2B ChatGPT apps can make money with far fewer users because the underlying contracts are much larger.

OpenAI has expanded app and plugin capabilities across Business, Enterprise and Edu workspaces, where administrators can control access, permissions and actions. ChatGPT can increasingly retrieve company information and perform work through connected systems rather than acting only as a general chatbot.

Suppose an internal ChatGPT workflow saves 200 employees ten minutes a day. Nobody needs to sell those employees a $5 app subscription. The company may happily pay thousands of dollars for the underlying software or implementation if the workflow saves enough time.

There is another business hiding here as well: building ChatGPT integrations for companies.

An agency or independent developer can charge businesses to connect internal databases, SaaS systems or operational processes to ChatGPT. Revenue comes from implementation, customization, maintenance and sometimes a broader software contract.

That business already works without waiting for OpenAI to build a developer revenue-sharing system.

For now, ChatGPT apps look much more mature as part of B2B software and integration work than as a mass-market paid consumer-app economy.

Does ChatGPT's checkout system solve the software monetization problem?

No. Instant Checkout solves a real commerce problem, while paid software still needs a more complete billing model.

The Agentic Commerce Protocol already handles a surprisingly sophisticated transaction flow. ChatGPT can help discover a product, pass the required information to the merchant and let the merchant process the order through existing systems. OpenAI says the merchant remains in control of fulfilment and the customer relationship.

That structure works naturally for groceries, clothing and other products because all the usual pieces already exist: item, price, inventory, payment processor and fulfilment.

Subscription software creates more questions.

Does a $15 purchase unlock one month of access or a fixed number of uses? Does the entitlement work on the developer's own website too? Can a company buy seats for employees? Who handles upgrades, cancellations and refunds? Does OpenAI eventually take a percentage?

OpenAI's current Apps SDK documentation still says broader monetization details are coming.

So Instant Checkout shows that OpenAI is serious about transactions. We still need another layer before charging for independent digital applications feels as ordinary as charging for an iPhone app.

How risky is it to build a business that depends entirely on ChatGPT?

Quite risky today. OpenAI controls discovery, permissions and access closely enough that a business built entirely inside ChatGPT carries much more platform risk than a normal standalone SaaS product.

OpenAI's developer terms are unusually clear on one point: developers receive no guarantee of placement, visibility, ranking or promotion. OpenAI can also decide whether an app is published or remains available.

The current transition from the original app directory toward Plugins makes the risk easier to see. OpenAI is still deciding how third-party capabilities should be packaged and discovered.

The same thing happened with custom GPTs. Developers built around the GPT Store, but OpenAI is now preparing to retire custom GPTs and migrate useful workflows into Plugins.

We should not read that as evidence that OpenAI is abandoning developers. The underlying ecosystem is actually becoming more capable. It does show that today's interface or directory structure may have a short shelf life.

The safer strategy is to own something outside ChatGPT: customers, proprietary data, an API, a brand, a website, a valuable workflow or a separate product.

Then ChatGPT can become a powerful channel without becoming a single point of failure.

Is there actually an early-mover advantage in ChatGPT apps?

Yes, but the useful advantage comes from learning what users do inside conversational software, rather than simply publishing first.

Traditional app stores reward history. Early apps can accumulate reviews, downloads and rankings that later competitors struggle to catch.

We do not yet know whether ChatGPT will produce the same kind of compounding advantage. OpenAI has said usage and user satisfaction can influence broader distribution, so apps collecting genuine engagement now could benefit later. The ranking system is still too opaque to put much value on that possibility.

The more concrete advantage is knowledge.

Developers building now can learn which prompts trigger their app, when users agree to connect an external service, where authentication kills conversion, which jobs people repeatedly perform and which workflows feel noticeably better inside a conversation than on a normal website.

Those lessons are difficult to copy from an Apps SDK tutorial.

By contrast, publishing dozens of thin apps around popular keywords looks much less interesting. The GPT Store already showed what happens when production becomes almost frictionless: supply explodes faster than meaningful demand.

Being early helps most when it gives us a year of customer behavior that the next developer does not have.

What is still missing before ChatGPT apps become a real App Store economy?

ChatGPT apps still need better discovery, easier software billing and much stronger developer analytics before we can treat the ecosystem like a mature app marketplace.

Discovery is the first gap. Developers need some confidence that building a good product can produce organic users. OpenAI currently gives no guarantee of ranking or visibility, and early builders have repeatedly complained about apps being difficult to find.

Billing comes next. External checkout is now useful, and commerce has progressed quickly, but developers still cannot rely on a standardized native subscription system for ordinary third-party software.

Analytics are surprisingly important too. Earlier this year, developers asked OpenAI for the kind of traffic and discovery information Apple and Google provide. OpenAI support acknowledged that ChatGPT did not offer equivalent visibility into traffic sources, rankings and app performance. Some builders said they had created their own dashboards simply to understand usage.

Those three gaps reinforce each other. A developer needs to know whether the app is being shown, whether people use it after seeing it and whether those users eventually become paying customers.

ChatGPT already has the hardest asset for any marketplace to create: an enormous audience actively explaining what it wants in natural language.

The commercial infrastructure around that audience still needs to catch up.

So, can ChatGPT apps make money yet?

Yes. ChatGPT apps can already make real money, but the strongest businesses today usually earn that money outside OpenAI's still-developing app marketplace.

Commerce businesses can turn conversations into purchases. SaaS companies can bring existing paid products into ChatGPT. B2B software can make expensive contracts more useful. Service businesses can generate valuable leads. Developers and agencies can also charge companies to build private integrations.

A much tougher case is the independent developer starting from zero with a small digital utility. ChatGPT still does not provide broad usage payouts, native software subscriptions remain unfinished, and organic discovery can be extremely weak for apps that receive no prominent placement.

The freshest changes make the direction clearer. OpenAI is consolidating apps and skills around Plugins, retiring custom GPTs, opening external checkout and continuing to build commerce. The ecosystem is moving toward businesses with real services behind them.

So the answer today is yes, with a fairly sharp dividing line.

If ChatGPT is a new interface or distribution channel for something you already know how to monetize, the opportunity is real now.

If you need ChatGPT itself to discover your app, deliver customers and handle the entire business model, we are still early.