An error I encountered while using Auto，We’re experiencing high demand for Composer 2 right now. Please switch to Auto, another model, or try again in a few moments.
Hey, this isn’t a Cursor bug. One of the models that Auto routed you to at that moment was temporarily out of capacity. The backend returns a rate limit, and you see that message.
What you can do right now:
Just resend the request. Auto will likely route you to a different model.
Or pick a model manually in the chat model selector Sonnet or GPT-5 or Gemini until the load drops.
About the error text, I agree. The wording switch to Auto is confusing when you’re already on Auto, and mentioning a specific model isn’t needed. This is a known UX issue, and the copy fix is already on the radar.
Good question. The model picker shows what’s available on your plan, but it doesn’t reflect real-time provider load. Capacity is checked when you send the request. If a specific model is at peak load at that moment, either on our inference or with an upstream provider like Anthropic, Google, or OpenAI, the backend returns RESOURCE_EXHAUSTED and you see the “high demand” message, even if the model looks normal in the dropdown.
What usually works in these moments:
Just resend, the spike often clears in a few seconds
Switch to another model, different providers cool down at different times
Use Auto, it routes to wherever there’s capacity
There isn’t a real-time load indicator in the picker yet, that’s valid UX feedback.
