Model availability in canada region with strict data residency

Question

Hi Team,

I require clarification regarding Azure OpenAI model availability in the Canada region.

Specifically, we need confirmation on the following:

Which Azure OpenAI models are currently available in the Canada region (e.g., Canada Central)?

Among those, which models guarantee strict Canadian data residency, meaning:

All inference processing occurs within Canada

  No cross-region inference or routing takes place
  
     No customer data is transmitted or processed outside Canada
     
     Are there any configurations, features, or deployment options (such as global endpoints or fallback routing) that should be explicitly avoided to ensure strict in-region processing?
     

I require this confirmation to ensure compliance with Canadian data residency requirements for our solution.

I would appreciate receiving the exact model names, deployment types and any relevant documentation references.

Thank you for your assistance.

Answer 1

Hi Pallavi Pundir

if you deploy Azure OpenAI in the Canada region (like Canada East or Canada Central), your data stays in Canada.

1. Azure OpenAI respects regional deployment. The data and inference stay within the selected Azure region.
2. If you choose Canada Central, both the compute (model inference) and data processing happen within Canadian data centers.
3. No cross-region data movement occurs unless you explicitly configure services (e.g., logging, backup) that write to other regions.

Azure OpenAI models available in Canada Central (regional availability):

Microsoft does not publish a single static list per region, but confirms that standard GPT families are available in Canada regions, subject to quota and access approval. The following model families are explicitly confirmed as deployable in Canada regions using regional deployments:

• GPT‑4 family (including GPT‑4 and GPT‑4 Turbo variants, subject to access approval)
• GPT‑4.1 family (gpt‑4.1, gpt‑4.1‑mini) as part of general Azure OpenAI regional availability
• GPT‑4o and GPT‑4o‑mini, only when deployed as regional (standard or provisioned) deployments, not global
• gpt‑35‑turbo (legacy but still regionally deployable)

For strict Canadian data residency compliance, Microsoft‑supported and documented configuration is:

• Azure OpenAI Service
• Region: Canada Central
• Deployment type:
  • Standard (Regional) or
    • Provisioned (Regional)
    • Endpoint: Regional endpoint (*.openai.azure.com)
• No global routing, no data zones, no preview/global features

Azure OpenAI Models and Regional Availability

https://learn.microsoft.com/en-us/azure/ai-foundry/responsible-ai/openai/data-privacy?tabs=azure-portal

https://learn.microsoft.com/en-us/azure/ai-foundry/foundry-models/concepts/deployment-types?view=foundry-classic

Helpful Links

• Azure OpenAI Standard Deployment Model Availability
• Azure OpenAI Service Documentation
• Azure Compliance Offerings
• Azure Subscription and Service Limits

Thank you.

Answer 2

In the Canada region, specifically Canada East and Canada Central, the following Azure OpenAI models are currently available:

• gpt-35-turbo-16k (0613)
• gpt-35-turbo (1106)
• gpt-4-32k (0613)
• gpt-4 (1106-preview)
• gpt-4 (0125-preview)
• gpt-4 (0613)
• gpt-4o (text-only implementation)
• gpt-4 (turbo-2024-04-09)

Regarding strict Canadian data residency, Microsoft ensures that for Azure OpenAI resources in the Canada region, all inference processing occurs within Canada, and no cross-region inference or routing takes place. Customer data is not transmitted or processed outside Canada without authorization.

To maintain strict in-region processing, it is advisable to avoid configurations or features that might involve global endpoints or fallback routing, as these could potentially lead to data being processed outside of Canada.

For further details, you may want to refer to the Azure OpenAI documentation that outlines model availability and data residency commitments.