Share via

AI foundry project Aurora weather model deployment issues

Kai Neuffer 0 Reputation points Microsoft Employee
2026-03-11T10:07:47.9766667+00:00

When I try to deplploy the aurora weather model as an AI Foundry project I get a very limited choice of VM SKUs. Only an NC_A100 serie is availble. This is different when I try to deply the model on an AI Hub project where are many more SKUs are available for this model, especially also the NC_T4 on which I have quota in the Sweden central region. Is there a way to fix this issue?

Foundry Tools
Foundry Tools

Formerly known as Azure AI Services or Azure Cognitive Services is a unified collection of prebuilt AI capabilities within the Microsoft Foundry platform

1 answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Manas Mohanty 16,105 Reputation points Microsoft External Staff Moderator
    2026-03-23T18:54:20.6966667+00:00

    Hi Kai Neuffer

    Each model asks specific VM type for deploying to endpoint.

    It is asking NC24ADSA100_V4 compute for deployment in this case.

    User's image

    You might need to raise quota if you want to serve the model as endpoint.

    Recommendation

    For POC Proposes, we should be able to call it via foundry client and run predictions.

    There are other varieties of Aurora mentioned here to register via Mlflow and deploy.

    Attached relevant document for reference

    https://github.com/microsoft/aurora/blob/main/package_mlflow.py

    https://mlflow.org/docs/latest/ml/deep-learning/transformers/large-models/

    Thank you.

    1 person found this answer helpful.

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.