Back to Blog
Growth5 min readSeptember 30, 2026

ChatGPT Users Can Now Spend Their Plan Inside Other Apps

OpenAI's DevDay opened ChatGPT Plus and Pro allowances to 16 partner apps, including Notion, Vercel and Devin. What 'Use your ChatGPT plan' means for pricing AI features in your SaaS.

Alex Rivera

Alex Rivera

Growth at NeedBase

At DevDay on 29 September, OpenAI turned Sign in with ChatGPT from a login button into a billing channel. ChatGPT Plus and Pro subscribers can now let 16 launch partner apps draw on their existing ChatGPT allowance, so the user pays OpenAI once and the partner app never needs its own API key or token bill for that user. Partners include Notion, Vercel, Devin, Warp and Amp, with Lovable listed as coming soon.

If you sell a product with AI features, this changes one of the hardest questions in your pricing: who pays for the tokens.

How it works

The login part is standard OAuth. According to Apidog, apps receive the user's name, email address and profile picture, never API keys or conversation history. Trending Topics notes that OpenAI in turn receives account IDs, email addresses, app details, IP addresses and device identifiers for authentication and fraud prevention, so OpenAI learns which tools its users rely on.

The new part is plan usage. When a user connects an app, eligible requests are deducted from their allowance for ChatGPT Work and Codex. Connecting an app adds no extra volume. Per VGTimes, users set a weekly cap for each app as a percentage of their allowance, and requests stop when the cap is hit. Separately purchased credits are only used with explicit consent, and there is no automatic fallback to the partner's own billing. Business and Enterprise workspace admins must approve the connection.

Who is in, and who is not

VGTimes lists the 16 launch partners. The commercial apps, available to Plus and Pro subscribers, are Amp Code, Conductor, Dactyl, Devin, Hermes Agent, Hyperagent, Kilo Code, Notion, Vercel, Vorflux and Warp. The open-source projects are OpenClaw, OpenCode, Pi and T3.

Access for everyone else is limited. Trending Topics reports that plan usage is currently open to open-source projects, locally running apps and selected partners, with commercial providers joining through a waitlist. OpenAI has said more partners will follow but has not given a timeline, and it has not announced a revenue share or payment terms of the kind app stores publish.

Why this matters for SaaS pricing

Most SaaS teams adding AI features pick one of three models: absorb the token cost in the subscription, sell credits, or ask users to paste their own API key. The first squeezes margin on heavy users. The second adds friction. The third works for developers and almost nobody else.

"Use your ChatGPT plan" is a fourth option: the user's existing subscription covers inference, and your app charges for everything else. For a tool whose main cost is tokens, such as a coding agent or a writing assistant, that could remove your largest variable cost for those users.

The trade-offs are real. You are routing your core feature through a competitor-adjacent platform that controls the caps, sees who your users are, and can change the terms. Only Plus and Pro subscribers can use commercial partner apps this way. And because usage stops when a user's weekly cap is reached, your product needs a clear fallback, or it simply stops working mid-task.

What to actually do

1. Work out your token cost per active user. Pull a month of usage and split it by customer. If a small share of users drives most of your inference bill, plan usage is a way to serve them without raising prices for everyone else.

2. Join the waitlist if tokens are your main cost. Commercial access is by waitlist. Getting in early costs nothing and tells you what the integration and terms actually look like.

3. Design for the cap. Decide now what happens when a connected user runs out: switch to your own metered credits, drop to a cheaper model on your own key, or show a clear message. Silent failure is the worst outcome.

4. Keep your own billing path. Treat ChatGPT plan usage as an optional payment method alongside your own plans, not a replacement. Anthropic, Google and xAI users are not covered, and Trending Topics reports that competing assistants are not participating.

5. Update your privacy notice. If you add Sign in with ChatGPT, say that OpenAI receives device and usage signals when users log in.

The bottom line

OpenAI is making the ChatGPT subscription a wallet that works inside other apps. For the 16 partners it removes the per-user inference bill for Plus and Pro users. For everyone else it is a waitlist and a pricing idea worth modelling now: if your AI costs are concentrated in heavy users, letting them bring their own plan could fix your margins. Just do not build a product that depends on it.

Found this useful?

Share it with a founder who needs it.

Ready to launch your product?

Join thousands of makers who launched on NeedBase.

Submit Your Product โ†’

Compare the Tools in This Post