getcursor
Support API keys for the Azure OpenAI Service
Hey, the Cursor team currently works directly with OpenAI to build the service, so I believe it would be unlikely for them to support Azure OpenAI any time soon. However, I'll leave your issue open as a potential future enhancement.
We could probably offer this! What sort of info would you need to specify? The API version, API base, and API key? Do your python scripts with azure look like this:
I will like that too, mostly because I had already setup this and I get it working
As of https://changelog.cursor.sh/#0233---azure-support this is implemented... Is there any plan to create documentation around this?
You can use Azure OpenAI by following these steps
- Click the settings icon on the top right
- Scroll down to Advanced and you will see such an option
I have configured the correct Azure API, but it does not work
Azure OpenAI Service can be used for the chat and edit. However, as @z4656207 pointed out, it is not yet compatible with the "New AI Project". It hangs at the Step 0.
I have configured the correct Azure API, but it does not work
Azure OpenAI Service can be used for the chat and edit. However, as @z4656207 pointed out, it is not yet compatible with the "New AI Project". It hangs at the Step 0.
Can confirm this is the case.
Another issue is that it seems that currently only one deployment/model is supported.
If I want to use gpt-4o via Azure for chat and gpt-4o-mini via Azure for in line editing it does not work I think.
s there way where we can set custom headers for Azure OpenAI in cursor models settings? headers like Ocp-Apim-Subscription-Key .. etc. for azure openai custom models deployments.
@truell20 if you have the time, it would be great to get a comment from the team on plans for further azure support.
It would be nice to add support for multiple deployment. For standard OpenAI a single api key let you use all models, for azure OpenAI we need to create a deployment for each model. Maybe using comma separated values in deploymentname field?
Also I'd like to know when I use chat when I'm using model from azure, it would be nice if the model name is called azure-{deploymentname}
@mallapraveen @alkampfergit Both of these are great suggestions, but could you possible open a new issue for them? I'll close this one for now as the original issue has been resolved, but as you've both said, could benefit from further development.