What to know first
Do not estimate included tasks from API token prices or treat promotional API credits as a Codex top-up. Fast mode and authorized third-party apps can consume existing allowance faster. Track actual payments, consumed usage, and remaining balances separately when budgeting.
Which kind of allowance are you looking at?
See the ChatGPT Pro comparison when selecting a subscription. Sharing here refers to Work and Codex; it does not establish one universal limit for every ChatGPT mode and feature.
| Category | Main purpose | Where to check |
|---|---|---|
| Included Work / Codex allowance | Eligible tasks with ChatGPT sign-in | Account usage and reset times |
| Purchased ChatGPT credits | Eligible usage beyond included limits | ChatGPT usage and billing |
| API balance or promotional API credits | Eligible API requests billed to the linked organization | API organization Billing and Usage |
Sources: OpenAI: ChatGPT Work and Codex pricingOpenAI: API credits for ChatGPT subscribers
Why do similar message counts consume different amounts?
Models, task complexity, context, reasoning, and tools affect consumption. Published message ranges are estimates, not fixed entitlements. Pro currently has no five-hour cap; check your account for other limits and resets.
For supported models, Fast consumes included allowance at 2.5x Standard and purchased credits at 2x. Neither ChatGPT multiplier should be applied to API-key billing.
| Billing resource | Relative to Standard on the same model | Boundary |
|---|---|---|
| Included allowance | 2.5x usage | Not a speed or completed-task guarantee |
| Purchased credits | 2x usage | Requires available credits and feature access |
| API key | API pricing applies | ChatGPT credit multipliers do not apply |
Sources: OpenAI: ChatGPT Work and Codex pricingOpenAI: Speed modes and usage
Why can another app consume your ChatGPT allowance?
In supported apps, eligible Plus and Pro subscribers can authorize Use your ChatGPT plan, drawing on existing Work / Codex usage. That permission is separate from signing in. Connecting an app adds no allowance, and the app can charge its own fees.
Settings > Usage shows app consumption and supports weekly percentage caps. A cap neither creates nor reserves an allowance for that app.
Sources: OpenAI: Sign in with ChatGPT
Can Pro or Business API credits extend Codex usage?
No. The benefit excludes Codex through API keys and does not increase ChatGPT or signed-in Codex limits.
Eligible paid Pro or Business subscriptions must be at least seven days old. An owner of both the subscription and API organization links them to claim the benefit. Credits expire with the applicable billing cycle without rolling over, and the first grant can be prorated. See the Pro and Business API credits guide for amounts and claiming steps.
What should you check when usage runs out?
This is AISub’s suggested troubleshooting order. Identify the limit before choosing a response so you do not fund a different platform and remain blocked.
For credit-based enterprise workspaces, consuming committed credits does not itself create a new invoice charge. Costs after the balance runs out depend on the contract and overage settings. A budget alert is not necessarily a hard spending cap.
| What you see | Check first | Then consider |
|---|---|---|
| A personal subscription usage limit | Reset time, model, speed, and background tasks | Waiting, extra credits, or a plan comparison |
| API credit_balance_exhausted | The request’s API organization and balance | Benefit eligibility, then API funding |
| A blocked company workspace task | Admin controls, credits, and overage settings | An administrator’s review of the agreement and limits |
Sources: OpenAI: ChatGPT Work and Codex pricingOpenAI: API credits for ChatGPT subscribersOpenAI: ChatGPT Work — usage and cost
How can you avoid counting credits twice?
AISub recommends two records: actual payments for cash budgeting, and usage plus balances for plan evaluation. Credits can be bought in one month and consumed in another. Do not add both the purchase and consumption of the same balance as separate cash payments.
All amounts below are hypothetical, not a bundled plan quote. Suppose you pay $20 for a subscription, $10 for extra ChatGPT credits, and a net $5 for the API after applicable benefits. Actual payments total $35. Unused promotional credits are not income and should not be subtracted from that total again.
- Read ChatGPT subscriptions vs the API to compare the app and developer interface.
- For API workload estimates, use the token cost calculation example with current model and tool rates.
Actual monthly payments = subscription payment + ChatGPT credit purchases + net API payments
Illustration: $20 + $10 + $5 = $35
Sources: OpenAI: ChatGPT Work — usage and costOpenAI: API credits for ChatGPT subscribersOpenAI: ChatGPT Work and Codex pricing
Frequently asked questions
Will adding API balance remove my Work subscription limit?
Not directly. Identify whether you are using ChatGPT sign-in or API-key billing; these are not one interchangeable balance.
Sources: OpenAI: ChatGPT Work and Codex pricingOpenAI: Speed modes and usage
Does connecting an app add another allowance?
No. Authorizing plan usage draws on existing limits. An app cap is a constraint, not additional or reserved capacity.
Sources: OpenAI: Sign in with ChatGPT
Is an enterprise dollar usage estimate an additional invoice?
Not necessarily. Committed credits, contracted rates, and overage settings affect charges. Compare planning estimates with the issued invoice for the same period.
Official sources and review notes
Sources checked on November 7, 2026. Rates and eligibility rules can change; confirm the latest official guidance and your own agreement.
- OpenAI: ChatGPT Work and Codex pricing
- OpenAI: Speed modes and usage
- OpenAI: API credits for ChatGPT subscribers
- OpenAI: Sign in with ChatGPT
- OpenAI: ChatGPT Work — usage and cost
This guide draws on the public sources listed below. Examples illustrate the calculation methods. AISub is an independent price index and is not affiliated with the product providers discussed.