ChiChieh HuangFOUNDER · AI ENGINEER

When You Can Take Your ChatGPT Subscription With You, How Does the AI Product Business Change?

Date2026.10.31
Length886 words
Reading~5 min
ChiChieh HuangFounder · AI engineer

Translated from the Chinese original · Read the original

LocatorLLM Plateau
2 wks
Overview

I watched the OpenAI DevDay announcements this time, and one new feature really hit home for me.

They announced that eligible ChatGPT Plus and Pro users can use their existing plan’s allowance inside supported third-party tools, like Devin, Notion and Vercel. As long as you authorize it, that model usage counts directly against your ChatGPT plan.

The reason this one landed so hard is that back when I was building side projects and AI products, I kept running into a really painful problem. A lot of users would ask: “I already pay for ChatGPT. Why should I pay for your product too?”

On the developer side, we know there’s a workflow and data integration behind the product, and that every API call is burning money. But honestly, from the user’s point of view it’s simple: I’ve already paid for AI once, and now that I’m using it somewhere else, I’m getting fleeced again 😭

To be clear, this doesn’t mean that once you pay for ChatGPT every other product becomes free. Third parties can of course still charge for their own software or services. The only thing users bring along is the AI model usage. Signing in and authorizing use of your allowance are two separate things.

But that difference matters a lot for people building products, because we finally get a chance to talk about “model cost” and “product value” separately. Users don’t necessarily have to buy another batch of model credits through my product, so I can focus my energy on convincing them of what problem this product actually solves for them. That line, “I already pay for ChatGPT,” used to be a reason to turn down yet another AI tool. Now there’s a good chance it becomes “well, if this tool can use the allowance I’ve already paid for, might as well give it a try.” I saw Simon Willison write in his live notes that he’d “been waiting for this for years,” and I totally get that reaction, because this really does deal with a very real friction between developers and users.

By the way, this also quietly gives people more reasons to keep paying for ChatGPT. Say I only use AI inside ChatGPT. When I’m deciding at the end of the month whether to renew, I’ll probably just look at how much I’ve used it lately. But if several tools I use every day can all draw from the same subscription, the way I judge that money changes. Even in a month when I barely open ChatGPT, I might stay subscribed because of the other tools. In effect, third parties are building good products while also helping deepen ChatGPT’s moat.

Devin has already shown how this works. Once eligible users connect their ChatGPT plan, the GPT model usage in their work counts toward ChatGPT, and usage of other models counts toward Devin. This doesn’t replace Devin’s paid plans. It lets the two subscriptions deliver value together.

My own feeling is that this will very likely affect which models developers pick in the future, too. When two models are about equally capable but one of them can use an allowance the user has already paid for, that’s one more reason for us to choose it. Competition is no longer just about capability and API pricing. It’s also about who users have already put their budget behind. There’s also the OpenAI Marketplace on the enterprise side, which is a similar idea: companies can spend their existing committed contract spend on partners’ software. It’s not quite the same as sharing a personal subscription, but it also lets existing budget flow to third parties.

Put simply, OpenAI is trying to build a new kind of relationship. Developers who plug into the platform get more than just models. They get a shot at a group of customers who already have an AI budget set aside, and the better those products are, the less willing users are to cancel their OpenAI subscription.

Still, you have to do the math. I don’t think this suddenly makes AI products easy to sell 😅 A shared allowance doesn’t mean a bigger allowance. Third parties are still consuming your existing plan’s usage. For heavy users, the question becomes: is the time this tool saves me worth eating into the allowance I would have used for other things? The product still has to prove what that software fee actually buys. Is it a few fewer rounds of copy and paste, or does it really run a tricky piece of work all the way through for you?

But the most interesting part this time is that OpenAI is giving third-party products’ growth a chance to be tied to its own subscription business. They don’t have to jump in and build every kind of application themselves to raise the value of their subscription through other people’s products.

What I’m more curious about next is whether, before buying an AI tool, people will increasingly ask: “Can this run directly on my ChatGPT subscription?” If that question starts to drive decisions, then OpenAI’s ecosystem really is more than just the logos on that partner list.

To friends building AI tools: would you rush to plug into this shared allowance feature? Share your thoughts in the comments. I’m also curious whether you think this will change your pricing strategy 🤔

#AI#BuildInPublic#AILogora#ChatGPT#OpenAI

End of the trail

886 words, and you made it to the end.

Newsletter

Get the next essay by email.

One email when a new essay goes up, nothing else. Unsubscribe in one click.

ChiChieh Huang
Surveyor

ChiChieh Huang

I build generative AI products and write about them, first in Chinese. Lately I’ve been researching agent memory and testing the ideas in Cairn.