If you write code in Cursor, Friday night just got messy.
OpenAI said it plans to stop shipping its models into Cursor, the AI coding tool SpaceX finished buying earlier this month for about $60 billion in stock. The proposed shutoff date is November 12, 2026. Reuters reports OpenAI is using a change-of-control clause in the Cursor deal to walk away.
The stated reason is trust. OpenAI wrote that it cannot be confident SpaceX will keep its technology inside the terms of service, citing past contract fights with Elon Musk's companies. It pointed at X, formerly Twitter, as an example of a breach after Musk bought that platform. Musk fired back on X that he "couldn't care less" and called Sam Altman and Greg Brockman untrustworthy.
Cursor co-founder Michael Truell, now a SpaceX executive, said the team is talking with OpenAI to try to fix this. Hours later Anthropic said it will pour more compute into Claude inside Cursor. Tom Brown, an Anthropic co-founder, made that public. So the short version for developers is this: GPT access through Cursor may go away in about three months, while Claude support looks like it is going the other way.
Here is the idea behind the drama. Coding agents sit on top of frontier models. When a lab owns the editor, rivals get nervous about distillation, ToS games, and who sees the traffic. We already watched a version of this when Anthropic cut Windsurf after OpenAI tried to buy that tool. Now OpenAI is cutting Cursor after SpaceX closed the deal. Same pattern, different lab.
Truell also said OpenAI models are only about 5% of Cursor user traffic right now. That softens the blow for most people. Still, if your workflow is built around GPT inside Cursor, start testing Claude, Gemini, or whatever else you already trust before November, not the week of the cutoff.
I do not love how personal this feud gets. But for day-to-day coding the practical move is boring and useful: keep your agent settings flexible, and do not bet your whole setup on one model vendor living forever inside one IDE.
