Requests Extremely Slow & Stuck, Mass Wasted Calls

Support Bug Reports
28

@Yang_Yichen - Thanks for the request ID.

I checked the backend logs, and the request shows the model returning empty responses repeatedly, which points to a provider-side issue rather than anything on your end. Which model were you using? And are you still seeing this, or has it improved in the last day or so?

For everyone in this thread: we had a service-side issue on April 3rd that caused elevated latency and stuck streams for some models. That was resolved, but some residual effects may have carried over into April 4th. If you’re still experiencing freezes or “Reconnecting…” today, please:

  1. Start a new chat (Ctrl/Cmd+N) rather than continuing a long conversation. We’re seeing that larger conversations are more prone to timeouts.

  2. Try switching to HTTP/1.1: Cursor Settings > Network > HTTP Compatibility Mode > HTTP/1.1. This avoids the HTTP/2 stream drops that cause the “Reconnecting…” state.

  3. Share a fresh request ID from a stuck request so we can check whether the current routing is healthy.

The persistent “Reconnecting…” state that doesn’t clear automatically is a known client-side bug that the team is actively working on.

29

New issue on version 2.6.22: I frequently get the message “Taking longer than expected…”.

Request ID:

  • 3c8cf791-d306-49f2-9504-d5bd6c34a0ed
  • 78748c24-d7a8-4f74-bbf5-ed442f0a76ad
  • f2d3a9d0-2f5c-4290-84e2-45154fb99bed
  • 44e6fcc6-d309-4761-b217-a4e788351f42

Model: GPT 5.2 extra high fast

HTTP Mode: HTTP/1.1

I have no idea how long I should wait, so I have no choice but to stop and retry. It’s really frustrating.

30

Please update to the latest version and see if the issue persists.