Skip to content
Cursor Handbook

API and API keys

Cursor issues API keys for its own platform endpoints, not for reselling model inference.

  • Extending Cursor
  • Varies by endpoint. Admin endpoints are Teams and Enterprise.

Checked against cursor.com on 25 September 2026

What it does

People search for a Cursor API key expecting one of two different things. The first is a key for Cursor's own API, which covers administrative and automation surfaces such as usage reporting, seat management and the AI code tracking endpoints that Enterprise plans include. Those keys are issued from the dashboard.

The second expectation is using a Cursor subscription as a general-purpose model API, calling Grok or Opus through Cursor from your own application. That is not what the subscription provides. Model access inside Cursor is tied to the editor, the CLI and cloud agents. For direct model calls, go to the model vendor.

The adjacent thing that does exist is bringing your own vendor key into Cursor, which routes requests through your account with that provider instead of Cursor's included usage. It changes who pays, not what Cursor can do.

Checked against API documentation on 25 September 2026.

Frequently asked questions

Where do I get a Cursor API key?
From the Cursor dashboard. Keys authenticate calls to Cursor's own API surfaces, such as usage and administrative endpoints, with the broader set available on Teams and Enterprise plans.
Can I use my Cursor subscription as a model API?
No. A Cursor subscription covers model use inside Cursor's own surfaces, the editor, the CLI and cloud agents. Calling a frontier model from your own application means buying access from that model's vendor.
Can I use my own OpenAI or Anthropic key in Cursor?
Cursor supports supplying your own provider key, which bills that usage to your account with the provider rather than drawing on Cursor's included allowance. Check the current docs for which surfaces honour a custom key.

Keep reading