Due to an error on the cursor server, I had to submit four requests to get a solution to a problem. Claude 4 sonnet-thinking is expensive, and even though I spent so many tokens, I received an interrupted and incomplete response.
I guarantee that my network is fine. In many requests, I did not receive the complete response from the AI model, yet I had to pay for cursor’s network error.
This issue previously existed in the Gemini 2.5 Pro model and is now occurring in the Claude model as well. Have you also encountered this situation?
Version: 1.4.5 (user setup)
Commit: 1.99.3
Date: af58d92614edb1f72bdd756615d131bf8dfa5290
Electron: 2025-08-13T02:08:56.371Z
ElectronBuildId: 34.5.8
Chromium: undefined
Node.js: 132.0.6834.210
V8: 20.19.1
OS: 13.2.152.41-electron.0