An Azure service that provides access to OpenAI’s GPT-3 models with enterprise capabilities.
Short answer: Visual Studio (MSDN) subscriptions aren't excluded from Azure OpenAI, but they sit in a lower quota category than pay-as-you-go or enterprise subscriptions. You can see this in the quotas and limits page, where several tables list "MSDN subscriptions" as their own column with much smaller limits than the other offer types. The same page also says increase requests are prioritised for customers who are actively using their existing quota, which is why a new MSDN subscription asking for more tends to get quick rejections.
What I'd check before sending more requests:
- In the Foundry portal, open the Quota page for your subscription and look at what's already allocated per model and region. Often there's a small default quota you can deploy with right away, without an increase.
- Try deploying one of the models that shows quota there, using the Global Standard deployment type, and keep the tokens-per-minute within what's listed.
- If you want to confirm which tier your subscription is in, the page has a
quotaTierscontrol plane call you can run with your own Azure CLI login. - If you need more than the MSDN allocation for real workloads, the realistic path is a pay-as-you-go subscription, since that offer type gets higher default limits.
The blog post you found may have been written when the defaults for that offer were different, so I wouldn't rely on it.
When you open the Quota page, do you see any non-zero allocation for the models you want, or is it zero everywhere?
If this answered your question, please consider accepting it as the answer. If the Quota page shows zero for every model, let me know and I'll be happy to keep helping.
Reference: