Feature request: Separate API keys for GPT/Whisper

[asifjahmed]

Hey @paul-gauthier! Really wonderful tool you've whipped up here! The concept of using git repos as the basis and making edits using diffs is truly out-of-the-box thinking, and I applaud you for this enormous contribution!

I have a small request, based on my use case currently where I am using a custom Azure instance of gpt-4-32k (Yes it is available if you set up an Azure account and request the access!!) yet the API keys do not necessarily work with other models. If gpt-3.5-turbo-16k were needed, for example, a different set of Azure parameters (namely 'deployment-id') would need to change. But in addition to that, Azure APIs do not include whisper - so the lovely voice features of aider do not work when using Azure API keys.

It would be lovely if the .aider.conf.yaml config file could support different configurable params per model, so that I can continue to use my own Azure setup for gpt-4-32k but otherwise use my OpenAI account for gpt-3.5-turbo-16k or whisper.

Best Regards, -asifjahmed

[olafgeibig]

I second this.

[olivMertens]

yeah Azure api include whisper in preview

[paul-gauthier]

Thanks for trying aider and filing this issue!

One approach might be to make a different config.yml file for each model, and then specify which config file to use via --config.

Not sure that will help with voice/whisper, if you need a different deployment id to connect to gpt versus whisper in azure. But it should make it easy to do aider --config gpt-4.yml and aider --config gpt-4-32k.yml, etc.

[paul-gauthier]

I think you can do this now since we've switched to litellm for connecting to models.

https://aider.chat/docs/llms.html

I'm going to close this issue for now, but feel free to add a comment here and I will re-open or file a new issue any time.

[loflet]

@paul-gauthier I am finding it difficult to configure aider with Azure OpenAI for voice. I have added the API key for Azure OpenAI and AzureAI in the env variable but still when I run /voice command, it records it but as soon as I press ENTER it throws error. Pasted the error's last line below

openai.NotFoundError: Error code: 404 - {'error': {'code': '404', 'message': 'Resource not found'}}

What am I missing here?