Connect OpenAI and track your API costs

Create an Admin API key, paste it into Quota Monitor, and save to track this month's API costs. Prepaid balances and ChatGPT subscriptions are not included.

Connect OpenAI to Quota Monitor

  1. Open Admin Keys, create a key and copy it

    Open the link below and select the OpenAI organization you use for API calls. An organization owner can create an Admin API key named Quota Monitor. A regular model API key will not work for this connection.

    Copy the full key when it is created; the full value is shown only once. If Admin Keys is unavailable, ask your organization owner to create it.

  2. Paste the key into Quota Monitor

    Open your project → Add connection → Add new connection, choose OpenAI, and paste the key into the Admin API Key field. Use any connection name, such as My OpenAI.

    OpenAI project ID is optional: leave it blank to view costs for the whole organization. Enter a proj_ ID only to track a specific OpenAI project. Do not enter your Quota Monitor project name here.

  3. Save and check the first sync

    Click Save and add to project, then wait for the first sync. This month's costs and a successful sync time confirm that costs were retrieved; saving the connection alone does not confirm a successful sync.

    You do not need to set a budget or enable a hard limit first. If you see a credential or permission error, check the key and organization using the questions below.

How to check OpenAI API usage and credits manually

  1. Open Settings › Organization › Billing to see the organization's prepaid credit balance and payment settings.
  2. Open the Usage page to review current-month API costs.
  3. Open the Limits page to compare organization and project monthly spend limits with the usage limit OpenAI assigns.
Open official OpenAI page(opens in a new tab)

What Quota Monitor reads from OpenAI

Costs only · Not a balance

Guide updated

Organization costs have been read successfully. OpenAI prepaid balance retrieval is not implemented. You can check prepaid funds on OpenAI's Billing page.

Required credentials
Admin API Key
Monitoring currency
USD · US dollar
What is monitored
Monthly costs / official limit or reference budget / corresponding difference · not account balance
Current-month costs from the organization Costs API, optionally filtered by OpenAI project ID. Without that ID, results may cover the entire organization.
  • Internal budget remaining is not an OpenAI prepaid balance. It cannot establish available funds; we do not infer account funds from spending.
  • Quota Monitor project IDs differ from OpenAI project IDs. ChatGPT subscriptions are separate from API account funds.
  • Limit amounts, organization fallback and reference budget display await live-account verification. A proj_ filter checks the project limit first. Only an explicit unconfigured response triggers organization fallback; blank selects the organization directly. A single Key ID filter disables broader cap comparison. Permission and endpoint failures never trigger fallback.
  • Headroom is the matching monthly limit minus reported costs, not real-time balance; reporting and enforcement can lag. Both organization limits and reference budgets use the same month's total organization costs, shared by all projects, while project costs remain the main amount. Failed organization cost reads retain the retrieved amount but show no difference.
  • When applicable project and organization limits are confirmed unconfigured, the highest organization monthly spend alert becomes the reference monthly budget. The API supplies the amount but no percentage; choosing the maximum is our default rule, not confirmation that it represents 100% or an official hard cap. You do not need to enable Enforce a hard limit or enter another amount. Missing or failed alerts do not produce an assumed budget. Provider tier caps and wallet funds remain separate.
  • The dashboard shows the latest successful cost snapshot. Refresh cannot remove upstream reporting delays. An old cost and timestamp after a failed query are not a new report.

OpenAI FAQ and troubleshooting

Can't find Admin Keys, or only have a regular API key?

Check Settings → Organization → Admin keys on the OpenAI API platform and select the correct organization. An organization owner must create the Admin API key; team members can ask their owner. A ChatGPT subscription does not provide this key.

Why is there no remaining balance after a successful sync?

This integration returns costs, not prepaid funds. Without a configured internal budget, no budget remaining is invented.

Why are my credentials rejected?

Use an unexpired, unrevoked Admin API key from the correct organization. A regular project model key cannot replace it.

Why are costs higher than expected for my project?

Check the OpenAI project ID. Without a provider project filter, the query may cover organization-wide costs.

Why does the reference budget differ from the console's monthly budget?

The reference budget uses the highest spend alert amount; the API provides no percentage. If that alert is not at 100% of the budget, the values can differ. Check Organization limits → Spend alerts. We do not modify alerts or enable hard enforcement.

Verify costs, limits and scope

After syncing costs, compare the same organization, project and UTC month. Check project caps at Project settings → Limits → Spend, or organization caps at Organization limits → Spend. Compare both the monthly amount and Enforce a hard limit state, not spend alert thresholds. If No project limit · Using organization limit appears, clear the project filter on the official costs page and compare total organization costs for the same UTC month. Organization headroom is not the organization cap minus just this project's costs. Check: project costs remain the main amount, with organization costs, cap and headroom labeled separately. Expand About spend limits to investigate read failures. Quota Monitor never changes provider caps, internal budgets or top-up settings for this.

Verify the reference budget when no hard limit is configured

If the card shows Reference monthly budget · From highest spend alert, open Organization limits for the same organization, find Spend alerts and compare the highest alert amount. Do not compare it with a hard cap or only this project's costs, and do not enable Enforce a hard limit just for monitoring. Check: the card identifies organization scope, the reference budget matches the highest spend alert, and the difference subtracts total organization costs for the same UTC month. Reaching the reference budget is a warning, not proof of blocked requests or exhausted funds. Sync again after changing or deleting official alerts. Missing or failed alert reads retain costs without reusing an old reference budget.

Why does the OpenAI API return a 429 billing or quota error?

Check error.code in the response. credit_balance_exhausted means the organization's prepaid credits are used up; add credits on the Billing page. organization_spend_limit_exceeded or project_spend_limit_exceeded means a monthly spend limit you set was reached; raise or remove it, or wait for the monthly reset. organization_usage_limit_exceeded means the usage limit OpenAI assigned was reached; request a higher limit. error.type can still be insufficient_quota, and retrying does not restore access.

Official OpenAI references

Prepared by Quota Monitor, not an official provider tutorial. Interfaces and permissions may change; check official pages. Verification is limited to the scope stated here.

Monitor OpenAI automatically

Checking a console by hand is easy to forget. Quota Monitor reads OpenAI through its official API for you.

  • Refreshes every 10 minutes by default, and you can refresh manually at any time.
  • Internal budget warnings, invalid credentials and sync failures rise to the top; reported costs are not wallet funds.
  • Group cost reports by project and read the same scoped costs through remote MCP. Internal budgets never change provider limits.
Start for free

See current OpenAI API prices