ChatGPT Pro and Business Now Include API Credits: Up to $500 a Month
If you pay for ChatGPT Pro or Business, since October 28 you have up to $500 a month in API credits waiting. And three conditions almost nobody read: a single linked organization, a prorated first grant, and expiry every billing cycle.
What the Benefit Actually Is
It isn't loose cash and it isn't a discount on the subscription. It works like this: you link your ChatGPT plan to an API organization you own, and every time the cycle renews OpenAI deposits into that organization credits equal to your plan's monthly amount. That balance is spent at normal API rates and is there to build apps and agents.
API Credits Bundled Into a Subscription, Not a Discount
The distinction matters because it changes the math. A discount lowers what you pay; this is a balance with an expiry date. If you don't use it, it disappears, and you still pay the subscription. OpenAI's docs are explicit: unused credits expire at the end of the billing cycle and don't roll over.
Live Since October 28, 2026, Pro and Business Only
The benefit shipped with the October 28, 2026 changelog. It covers ChatGPT Pro and ChatGPT Business. It does not cover Plus, Go or Free, and ChatGPT Enterprise is out: if your plan moves to Enterprise, the API organization link is removed at the end of the billing cycle. Free trials don't count either — your paid subscription has to be at least seven days old.
Read also
How Much You Get Per Plan
Pro 100, Pro 200 and Pro 500: $100, $200 and $500 a Month
All three Pro tiers receive credits matching their subscription amount: $100, $200 and $500 in API credits each cycle. The $500 tier launched at DevDay in late September, as we covered in our DevDay 2026 piece.
Business: $20 per Standard Seat, $100 per Premium Seat, Capped at $500
On Business the math runs per seat: $20 for each paid Standard seat and $100 for each paid Premium seat. All seat credits pool into one workspace balance capped at $500 per month. On annual plans paid upfront, the whole year's grant arrives in one issuance and expires at the end of the annual cycle, with a $6,000 cap.
Why the $500 Cap Is the Part People Get Wrong
Ten Premium seats don't give you $1,000: they give you $500, because the cap belongs to the workspace, not the seat. Concrete example: three Standard seats plus two Premium seats total $260 a month. Adding one more Premium takes the next cycle to $360. The pool is recalculated from the seats on the plan when the credits are deposited.
How to Claim Them, Step by Step
The official flow has five steps and you can start from either side: ChatGPT or the API platform.
Link the Subscription to an API Organization
From ChatGPT: sign in with the account that owns the subscription, open Settings and Billing (on Business, the workspace billing settings or the Admin Console). Find the API credits benefit and start the claim flow. Review which API organization will receive the credits and confirm. From the API platform the path runs the other way: pick the organization, open Billing and select the subscription you want to link.
Requirements people discover halfway through: you must own the ChatGPT subscription or workspace and also own the API organization, and use the same email address on both accounts.
The First Grant Arrives Prorated
The first deposit isn't the full amount: it is prorated for the time left in your current billing cycle, and it expires at the end of that same cycle. Subscribe on the 28th and your first month is almost symbolic. The following ones arrive in full.
What Happens When the Cycle Renews
The balance lands automatically on each renewal for as long as the benefit exists and your plan is eligible. You activate it once. If you need to change the receiving organization, there's no button for it: you contact support. Choosing well the first time saves you the ticket.
The Three Footnotes That Decide Whether It Fits
They Expire Every Cycle and Don't Roll Over
Practical rule: if your monthly API spend doesn't reach the plan's amount, the leftover evaporates. There's no cushion for next month. One point in your favour: free credits are consumed before any you've purchased, so your paid balance stays untouched while the benefit lasts.
They Don't Raise Your ChatGPT or Signed-In Codex Limits
The docs say it plainly: the benefit doesn't increase your ChatGPT usage allowance or your signed-in Codex allowance, nor your personal usage-credit balance, nor the Business workspace credit balance. ChatGPT billing and API billing remain two separate worlds.
Codex Through API Keys Is Excluded
This is the detail that will bite anyone automating code. The credits cover API Platform usage billed to the linked organization, but Codex usage through API keys is not supported by the benefit. If your plan was to pay for the agent writing your code with these $500, that door is closed.
Running the Numbers
If You Already Spend on the API, It's Real Money
If your API bill is around or above your plan's amount, the benefit is cash in hand: with a Pro 500 where you were already spending $500 on the API, the $500 ChatGPT is effectively free. It's the same logic behind September's price war, when Anthropic and OpenAI cut prices on the same day: the subscription and the consumption compete for the same wallet.
If You Spend Less, the Rest Evaporates
Spend $80 a month on the API while paying for Pro 200 and the real benefit is $80, not $200. The rest is a gift. The right question isn't "how much do they give me" but "how much do I actually spend each month".
If You Only Use ChatGPT, Don't Upgrade For This
If you never touch the API, the credits do nothing for you and you shouldn't move up a tier for them. Same if you use Codex inside ChatGPT: the benefit doesn't touch that limit.
Anthropic Did the Same Three Weeks Earlier
Claude Max and Ultra: the Same Amounts
Anthropic went first earlier in October: Max plans get $100 a month in Claude Platform credits and Ultra gets $200. On Team it's $20 per Standard seat and $100 per Premium seat, pooled with a $500 monthly cap. Same pattern: they expire at the end of each cycle, don't roll over, and are spent before purchased credits.
The Difference: Claude's Balance Belongs to the Whole Organization
With Claude, credits are shared by everyone holding an API key in the linked organization and don't cover interactive Claude Code or usage on Bedrock, Vertex AI or Foundry. With OpenAI, the balance belongs to an organization you designate and is spent with project keys from that organization. Simon Willison, who keeps subscriptions with both companies, summed it up on October 29 as "pretty much exactly the same deal".
What It Means If You Build on the API
Prototypes and Low-Volume Agents Inside the Plan
For demos, proofs of concept and low-volume agents, marginal cost drops to zero inside a plan you were already paying for. That's exactly the range where paying twice hurts most. For a production product with real traffic, you still need your own billing: the balance is an advance, not a budget.
What the Credits Actually Cover
The official list includes the Responses API for text, code and image understanding, the Agents API with its hosted sandbox charges, the Decisions API, GPT-Live and the Realtime API for voice, the Images API, and built-in tools with their storage, such as web search, file search and Code Interpreter.
When They Run Out, the Card Gets Charged
The organization spends free credits first. Once the balance hits zero with no purchased credits left, requests fail with an HTTP 429 insufficient-credit error. To keep going you add paid credits or turn on Auto Recharge. If your agent runs 24/7, that moment arrives sooner than it looks. Other low-price providers, like Moonshot's Kimi K3.1 Pro, remain the alternative when volume is what matters.
Conclusion
The benefit is real, but it's an expiring balance, not a discount. If your API spend doesn't reach the plan's amount every month, what you're buying is headroom, not savings. The smart move is to check your last three invoices before picking a tier: if the number matches the plan, ChatGPT is genuinely free. This blog keeps measuring what building with AI actually costs, not what the price tag says.
