Azure GPT-5 API Key Not Working – Internal Server Error

Describe the Bug

Hello,
I am encountering an error when using the GPT-5 model via Azure. The API key fails with an internal server error.

Request ID: 21abad18-2485-4c67-bc8b-62ea3335dcb1

Please assist in resolving this issue.

Thank you.

Steps to Reproduce

Open the application

Attempt to connect to GPT-5 model using Azure OpenAI API.

Provide a valid API key and endpoint.

Send a request to the GPT-5 model.

Observe that the request fails with an internal server error.

Operating System

Windows 10/11

Current Cursor Version (Menu → About Cursor → Copy)

Version: 1.4.3 (user setup)
VSCode Version: 1.99.3
Commit: e50823e9ded15fddfd743c7122b4724130c25df0
Date: 2025-08-08T17:34:53.060Z
Electron: 34.5.1
Chromium: 132.0.6834.210
Node.js: 20.19.0
V8: 13.2.152.41-electron.0
OS: Windows_NT x64 10.0.26100

Does this stop you from using Cursor

Yes - Cursor is unusable

2 Likes

Hey, thanks for the report. Could you please confirm that other models from the Azure API are working without any issues?

GPT-4o works without issues, but on Cursor it shows “errored, not charged.” GPT-5, on the other hand, throws an error right away and doesn’t work, as I mentioned above. Any solution?

I’m not entirely sure, but what is the model called in the Azure interface? It might have a slightly different name. In that case, you need to add this model manually. If its name is exactly the same, I’ll pass this on to the team for further investigation.

its exact name is gpt-5, there’s no difference at all. there’s also a model called gpt-5-chat, which i tried as well by adding it as a new model, but it gives the same error and doesn’t work either…

1 Like

I’m getting an error as well:
Request ID: 52ba9653-b9e0-49d6-99a2-b7c132fa18e6
{“error”:“ERROR_UNSPECIFIED”,“details”:{“title”:“Internal server error.”,“detail”:“If the problem persists, please check your internet connection or VPN, or email us at [email protected].”,“additionalInfo”:{},“buttons”:,“planChoices”:},“isExpected”:false}
ConnectError: [internal] Error
at mel.$endAiConnectTransportReportError (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:4918:251962)

I’m using gpt-5 using the API key (model is hosted on Azure).
Using the exact same name for the model (added manually) as hosted: gpt-5-xxxx
If I switch to GPT-4.1 (also on azure) it works fine.

1 Like

I’ve just noticed that turning on Azure OpenAI API key for my AOI GPT-5 deployments breaks a manual model selection.
With it enabled I can only get a response if I select “Auto” which seems to default to sonnet 4. If I manually select Sonnet 4 with AOI key for GPT-5, it doesn’t work.
I get a generic “if the problem persists please check your internet connection…”

1 Like

I have the same error:If the problem persists, please check your internet connection or VPN, or email us at [email protected].

(Request ID: 8849ace8-605b-4fd7-bb63-6b8a70db5c32) But I switch to gpt4.1,it works

1 Like

The issue still hasn’t been resolved, and there has been no progress. I don’t want to open a new ticket for the same problem. The trial period for ChatGPT-5 has ended, and we still cannot access it. Many other users are experiencing the same issue. I kindly request that this be resolved as soon as possible.

I have the exact same issue. I am using a custom name

@Broqee @Gunwoo_Lee we are working on an adjustment and are tracking all reports of this issue to see impact and different use cases.

Please be as detailed as you can if you see additional info.

Thanks, I expect you to figure it out soon. That’s really important to me..

Same issue, using name “gpt-5”

Request ID: 9724791b-fc5b-4394-bf13-47bb43b57ac1
{“error”:“ERROR_UNSPECIFIED”,“details”:{“title”:“Internal server error.”,“detail”:“If the problem persists, please check your internet connection or VPN, or email us at [email protected].”,“additionalInfo”:{},“buttons”:,“planChoices”:},“isExpected”:false}
ConnectError: [internal] Error
at yel.$endAiConnectTransportReportError (vscode-file://vscode-app/Applications/Cursor.app/Contents/Resources/app/out/vs/workbench/workbench.desktop.main.js:4918:251962)

Describe the Bug

Unable to verify OpenAI API Key.

In the attached video you can see the process, it looks that the issue in the request body that Cursor sends, API returns 400 because of the deprecated field “max_tokens”

Steps to Reproduce

  1. Go to the models
  2. Open “API Keys” collaps
  3. Past your key in the open AI Key input and click verify

Actual results: nothing happens, UI still presents “Verify” button after loading.

Expected Behavior

Some confirmation about that the token is valid and ready to use

Screenshots / Screen Recordings

Operating System

Windows 10/11

Current Cursor Version (Menu → About Cursor → Copy)

Version: 1.4.5 (user setup)
VSCode Version: 1.99.3
Commit: af58d92614edb1f72bdd756615d131bf8dfa5290
Date: 2025-08-13T02:08:56.371Z
Electron: 34.5.8
Chromium: 132.0.6834.210
Node.js: 20.19.1
V8: 13.2.152.41-electron.0
OS: Windows_NT x64 10.0.26100

Does this stop you from using Cursor

Yes - Cursor is unusable

Same error here:

”If the problem persists, please check your internet connection or VPN, or email us at [email protected].”

Request ID: a3d33c31-786b-4129-b1aa-de77caf1e47a
{“error”:“ERROR_UNSPECIFIED”,“details”:{“title”:“Internal server error.”,“detail”:“If the problem persists, please check your internet connection or VPN, or email us at [email protected].”,“additionalInfo”:{},“buttons”:,“planChoices”:},“isExpected”:false}

Any update? Many people suffer from this problem.

Not yet the team is working on it and we will let everyone know once we have a solution.

1 Like

I’m facing the same issue, i also see no option to set the reasoning, verbosity, input and output token limit for models like GPT-5 when setting it up using Azure AI Foundry or Azure API key.

No results? It shouldn’t be this difficult!

@Broqee team is already looking into API key integration and improvements required there. As we handle several API providers its not always a quick fix due to compatibility requirements.