AI NewsDev toolsReported

OpenAI will stop serving its models in Cursor after the SpaceX purchase

OpenAI has set 12 November 2026 as the date it stops serving its models inside Cursor, after SpaceX bought the editor. Cursor says OpenAI models cover 5 percent of its customers.

AI News

Editorial2 min read

LinkedInX
The OpenAI logo

Image: Engadget

Why it mattersIf your team picked a coding tool because of the model behind it, that link can be ended by a third party without your input, and a date on a calendar is all the warning you get.

OpenAI has said it will stop serving its models inside Cursor on 12 November 2026, following SpaceX's acquisition of the editor. Engadget reported the decision on 28 August, and OpenAI published its reasoning the same day in a post titled "Our decision on Cursor following its acquisition by SpaceX".

According to Engadget, OpenAI's stated reason is that it cannot be confident SpaceX will use the technology within its terms of service, and it points to what it calls its experience with Elon Musk's companies violating contracts.

Cursor keeps working after that date. It also serves models from Anthropic, Google and SpaceXAI. Cursor co-founder and chief executive Michael Truell wrote on X that OpenAI models serve 5 percent of the tool's customers, and that the team is in talks with OpenAI. Anthropic's Tom Brown said his company will keep increasing compute to support Claude models in Cursor.

The part that is unusual

A model provider ending an agreement is ordinary. Naming a date more than two months out, in public, and giving the reason as a dispute with the buyer is not.

For a team using Cursor, none of the normal options apply here. You cannot fix this with a contract, a bigger plan, or a support ticket. Two companies you do not have a relationship with decided something between themselves, and the tool you use changed.

What to do before 12 November

If your team runs Cursor on OpenAI models, you have about ten weeks. Three things are worth doing in that time, and none of them takes long.

Find out which model your team actually uses. In most teams this was set once and never revisited, so the honest answer is usually "whichever one was default". If that model is from OpenAI, you are in the 5 percent.

Try the same work on Claude or Gemini inside Cursor now, while the old option still exists. A model swap changes how an assistant behaves on your codebase in ways a benchmark does not predict. Finding that out in November, under time pressure, is the costly way to learn it.

Write down what you would do if this happened again with a different pair of companies. That is the real lesson here. The question is not whether Cursor is a good editor, it is how much of your workflow depends on a commercial agreement between two other parties.

Teams that keep their prompts, their review rules and their test suites outside the editor have an easier time with this. Anything kept only inside one vendor's product changes when that product changes.

Source

This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.

Share
LinkedInX
Start a project