Stuck in planning-next-moves

Support Help
,
1

Title:

Cursor chat stuck on “planning next moves” and fails with ERROR_OPENAI; only Auto mode works (macOS)

Version:

Version: 2.3.30
VSCode Version: 1.105.1
Commit: d1289018cc3fcc395487f65455e31651734308d0
Date: 2026-01-08T05:21:38.211Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 23.4.0

Only auto works , fail to use any s

Request ID: 0bf64944-be79-4a71-902d-c2b782b35ab1
{“error”:“ERROR_OPENAI”,“details”:{“title”:“Unable to reach the model provider”,“detail”:“We encountered an issue when using your API key: Provider was unable to process your request\n\nAPI Error:\n\n\nError\n”,“additionalInfo”:{},“buttons”:,“planChoices”:},“isExpected”:true}

Description:
In Cursor, chat suddenly stopped working normally. I can only use **Auto mode**. Any selected model—both built-in models (e.g., **Sonnet 4.5**) and my custom model (**GLM-4.7**)—fails to respond and stays on **“planning next moves”**, then returns a request error.

Error message:
Request ID: `0bf64944-be79-4a71-902d-c2b782b35ab1`

{“error”:“ERROR_OPENAI”,“details”:{“title”:“Unable to reach the model provider”,“detail”:“We encountered an issue when using your API key: Provider was unable to process your request\\n\\nAPI Error:\\n\\n\`\`\`\\nError\\n\`\`\`”,“additionalInfo”:{},“buttons”:\[\],“planChoices”:\[\]},“isExpected”:true}

Steps to reproduce:

  1. Open Cursor.
  2. Open Chat.
  3. Select a specific model (e.g., Sonnet 4.5) or a custom model (GLM-4.7).
  4. Send any message.

Expected result:
The selected model should respond normally.

Actual result:
The chat gets stuck on “planning next moves” and then errors with `ERROR_OPENAI`. Only Auto mode works.

Troubleshooting attempted:
I completely cleared Cursor data, uninstalled/reinstalled, and ran:

rm -rf \~/Library/Application\\ Support/Cursor
rm -rf \~/Library/Caches/Cursor
rm -rf \~/Library/Preferences/com.cursor.app.\*
rm -rf \~/Library/Saved\\ Application\\ State/com.cursor.app.savedState

Additional note:
After rebooting, the selected model works only for the **first** chat message. After that, the issue returns (stuck on “planning next moves” + request error).

3

Hey, thanks for the report.

This is a known issue. When the Override OpenAI Base URL setting is enabled for GLM-4.7, it affects all models, including Cursor’s built-in models (Sonnet 4.5 and others). Cursor then tries to route its own models through your custom Base URL instead of its provider, which leads to the ERROR_OPENAI error.

Workaround:

  • Turn off Override OpenAI Base URL when you want to use Cursor’s built-in models (Sonnet 4.5, etc.)
  • Turn it back on only when you need GLM-4.7
  • Manually switch the setting depending on the model you’re using

Important: GLM isn’t in Cursor’s list of officially supported providers (only OpenAI, Anthropic, Google AI, Azure OpenAI, AWS Bedrock). If you use it via Override Base URL, you may see unstable behavior, and the GLM API can lose reasoning content between messages, which reduces answer quality.

Similar case: Cursor Models Fail When Using BYOK OpenAI Key with Overridden Base URL (GLM-4.7)

Let me know if this helps.

4

Thanks, your solution works.