Issue when using custom OpenAI Key with GPT 5 High Fast

Support Bug Reports
1

Describe the Bug

An issue seems to have started happening recently (a couple of releases ago). Now, when using a custom OpenAI key and the GPT-5 High Fast model, I’m getting an error about “prompt_cache_retention is not supported on this model”. See below the full error:

Request ID: 044f0db9-44e4-40fc-a0a9-ec68448e75af
{"error":"ERROR_OPENAI","details":{"title":"Unable to reach the model provider","detail":"We encountered an issue when using your API key: prompt_cache_retention is not supported on this model\n\nAPI Error:\n\n```\n{\"type\":\"error\",\"sequence_number\":1,\"error\":{\"type\":\"invalid_request_error\",\"code\":\"invalid_parameter\",\"message\":\"prompt_cache_retention is not supported on this model\",\"param\":\"prompt_cache_retention\"}}\n```","additionalInfo":{},"buttons":[],"planChoices":[]},"isExpected":true}
ConnectError: [invalid_argument] Error
    at iJc.$endAiConnectTransportReportError (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:4989:399337)
    at Ewo._doInvokeHandler (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:36007)
    at Ewo._invokeHandler (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:35749)
    at Ewo._receiveRequest (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:34514)
    at Ewo._receiveOneMessage (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:33336)
    at $Dt.value (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:492:31429)
    at _e._deliver (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:49:2962)
    at _e.fire (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:49:3283)
    at ugt.fire (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:4974:12156)
    at MessagePort.<anonymous> (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:7497:18439)

Steps to Reproduce

Add a custom OpenAI API key, try using the GPT-5 High Fast in a custom mode.

Expected Behavior

Requests work.

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.0.64 (Universal)
VSCode Version: 1.99.3
Commit: 25412918da7e74b2686b25d62da1f01cfcd27680

For AI issues: which model did you use?

Model name: GPT-5 High Fast

For AI issues: add Request ID with privacy disabled

044f0db9-44e4-40fc-a0a9-ec68448e75af

Does this stop you from using Cursor

Yes - Cursor is unusable

1 Like
Custom API key breaks other models
`Error: Unable to contact model provider` for GPT-5 (non-codex) family of models
2

Hey, thanks for the report. This is a known bug affecting GPT-5 High/Fast when using custom OpenAI API keys. Cursor is incorrectly sending the prompt_cache_retention parameter, which those models don’t support.

Temporary workaround: use GPT-5 Codex instead of GPT-5 High/Fast with your custom API key - it’s confirmed to work (see the related report: Custom API key breaks other models).

I’ll pass this to the team. Please let me know if the workaround resolves the issue.

4

Hi, any updates on this? Unfortunately, the workaround does not work for our use case. Any idea of where this issue lands on the team’s roadmap?
Thank you.

2 Likes
5

This topic was automatically closed 22 days after the last reply. New replies are no longer allowed.