Using your own OpenAI API key in Cursor: what it bills and how to cap it

Checked November 9, 2026. Dates are proposed, not confirmed. Re-check the sources below.

The short answer

OpenAI says it plans to wind down its contract supplying OpenAI models to Cursor and has proposed November 12, 2026 as the transition date. The date is not confirmed, and Cursor may end access sooner. If you add your own OpenAI API key, you keep OpenAI models in Cursor's local Chat and Agent, billed to your OpenAI API account at API prices, separate from any Cursor or ChatGPT subscription. [1] [2]

What OpenAI said

OpenAI notified SpaceX that it intends to wind down the contract, with a proposed shutoff date of November 12, 2026. [1]

The proposed transition period is November 12, 2026. Cursor may choose to end access sooner. The official termination date is to be shared once both companies confirm it. [2]

What your own key covers

OpenAI's help article says your own key covers local Chat and Agent in the Cursor desktop app. That usage is billed at OpenAI API prices on your OpenAI API account. A ChatGPT subscription does not include API usage. [2]

The same article says your own key does not cover:

• Cursor Tab or autocomplete
• Auto model selection
• Cloud or Background Agents
• Automations
• Cursor CLI
• Cursor API or SDK
• Cursor-provided models

OpenAI's setup steps are: [2]

1. Create an OpenAI API key on an API account with billing configured.
2. Open Cursor Settings, then Models, paste the key, and save or verify it.
3. Enable the OpenAI model.
4. Select that model in Chat or Agent.

Cursor's API keys page says the provider bills you directly. On individual plans, requests made with your own key do not draw from included usage. On Teams and Enterprise plans, the Cursor Token Rate of $0.25 per million tokens can still apply, with exemptions. [3]

OpenAI also lists the Codex IDE extension and AI gateway providers as other options. [2]

Why the bill can surprise you

Usage is billed per request to your OpenAI account, so agent runs add up on the key. Available models and usage limits depend on your OpenAI organization. [2] The runaway agent checklist is the loop side of that. This page does not invent a dollar example. The bill estimator is where to try a request and token count.

How to cap it

Create a dedicated OpenAI project and key for Cursor so its spend is separable. See the per-key spend guide. OpenAI's documented spend cap is monthly, per organization or per project. The hard spend limit guide is that monthly cap. Turning auto-reload off is in the auto-reload guide. The same trap with an Anthropic key is the Claude Code billing guide.

Where BillCurfew fits

BillCurfew watches OpenAI, Anthropic, and OpenRouter API spend. It polls provider usage about every 5 minutes and does not proxy traffic, so there is no code change and prompts never reach it. Polling is not instant. You can set a daily cap or a spike alert per key, and alerts go by email. Free is $0 for one provider and one key, with email alerts and no card. A manual kill works on every plan, including Free. Auto-kill is paid-only and stays off until you turn it on. OpenAI key deletion is permanent. Solo is $19, Team is $49, and Agency is $149 per month, each with a 14-day trial. Start free. BillCurfew watches the OpenAI key's spend, not Cursor itself.

FAQ

Sources

Re-read before Nov 12.

1. OpenAI, Our decision on Cursor following its acquisition by SpaceX. https://openai.com/index/our-decision-on-cursor-following-its-acquisition-by-spacex/
2. OpenAI Help Center, Using OpenAI models in Cursor. https://help.openai.com/en/articles/20001506-using-openai-models-in-cursor
3. Cursor, API keys. https://cursor.com/help/models-and-usage/api-keys