@github-actions[bot] commented on GitHub (Nov 8, 2025):
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.…
@NickM-27 commented on GitHub (Oct 6, 2025):
I don't think another port makes sense, this will greatly complicate the nginx config management and we have already implemented custom roles.
You…
@StuartBJustice commented on GitHub (Oct 17, 2025):
Can we get the docs updated then please as they currently state: 'The model field is not required in your configuration as the model is part of…
@putt6359 commented on GitHub (Nov 13, 2025):
It would be interesting to see if ROCm can use it directly or not. The main problem with adding support is that we don't have any devices with NPUs…
@hawkeye217 commented on GitHub (Oct 17, 2025):
None of the other devs and I have the ability to test that, so @vfc2 will have to confirm before we update the docs.
@vfc2 commented on GitHub (Oct 17, 2025):
@hawkeye217 For context:
Pre-GPT5, the model name is part of the endpoint URI:
`https://instance.cognitiveservices.azure.com/openai/deployments/gpt-4.1-…
@NickM-27 commented on GitHub (Feb 6, 2026):
A user has successfully run with the NPU https://github.com/blakeblackshear/frigate/discussions/21906#discussioncomment-15722409
@hawkeye217 commented on GitHub (Oct 17, 2025):
Thanks @vfc2.
So pre-GPT5, if the model is specified in the config and is also specified as part of the endpoint, will that cause issues?
It…
@vfc2 commented on GitHub (Oct 17, 2025):
Good point @hawkeye217, I've cloned the repo to test - The model in the endpoint is not being used. If the model field is not provided in the config, it…
@hawkeye217 commented on GitHub (Oct 17, 2025):
Great, I'll update the docs to just make the model field mandatory. Thanks for your help!
@vfc2 commented on GitHub (Oct 17, 2025):
You're welcome, thanks for your work on frigate. Happy to help for future work related to Azure.