After yesterday’s crash (July 31), I have it complaining on any model and asking to switch to Auto.
Hey, this is simpler than it looks. It’s not related to the July 31 incident, it’s the expected behavior when you hit your usage limits.
Once the included plan usage runs out, named models Claude, GPT, etc stop being available and Cursor will suggest switching to Auto. Auto keeps working, it might be a bit slower sometimes due to the queue. That’s exactly what you’re seeing: any named model errors out, but Auto goes through.
Named models will be available again when:
- your limit window resets, or
- you enable or increase on-demand usage, or
- you move to a plan with more included usage.
You can check your current limits at cursor.com/dashboard. If you have questions about the exact usage numbers, email [email protected] and the team can look into your account details.
The July 31 incident is already resolved and isn’t related to this behavior. Let me know if anything’s still unclear.
I can say with confidence that on the 31st you deployed updates that took down your entire infrastructure. Among those changes was the routing layer, along with updated rules governing model behavior and routing.
How do I know? Let’s just say… I know. 
Given the direction Cursor is heading, and where the product seems to be going, I’ve decided to move to one of your competitors. As someone who has supported Cursor almost since its launch, it wasn’t an easy decision. I’ll still keep a minimal part of my workflow on Cursor, but the majority of my work is moving elsewhere.