Azure OpenAI error

Hello Bikeh,

Welcome to Microsoft Q&A and Thanks for reaching out,

The 403 Forbidden is the key signal here. This means your request is reaching Azure OpenAI Service, but the service is rejecting it due to authentication, permissions, or configuration issues (not a code/SDK problem).

Also, from your logs Streaming error: Forbidden, falling back… followed by another 403

This confirms it’s not a streaming issue it’s a core access/config issue.

Steps to resolve

1. Verify authentication.

Ensure your endpoint URL is correct:

https://<your-resource-name>.openai.azure.com/

Make sure the API key matches the same resource

Confirm your client (Python/JS/HTTPX) is actually sending:

• api-key header OR
• correct Entra ID token

Endpoint and key mismatch are common cause of 403

2. Check deployment name

Azure uses deployment names, not raw model names.

• Go to Azure Portal → OpenAI → Deployments
• Verify exact name (case-sensitive)

3. Confirm you have access (RBAC / IAM)

Go to Azure Portal → OpenAI resource → Access control (IAM)

Ensure your user/service principal has:

• Cognitive Services OpenAI Contributor
• or Contributor / Owner

4. Check if model is deployed & allowed

• Make sure the model is actually deployed in your resource
• Some subscriptions/regions don’t have access to all models

5. Review networking restrictions

If your resource has Firewall rules, Private endpoint, Selected networks only

Then your request may be blocked.

Check Networking → Public access enabled?

Your IP/VNet is allowed?

6. Confirm subscription access

If this is a new setup Azure OpenAI access must be approved first

If not approved, then all calls return 403

7. Check service health

Look at Azure Service Health / Resource Health

Ensure no outage in your region

Try a simple non-streaming call:

from openai import AzureOpenAI
client = AzureOpenAI(
    api_key="YOUR_KEY",
    azure_endpoint="YOUR_ENDPOINT",
    api_version="2024-02-15-preview"
)
response = client.chat.completions.create(
    model="YOUR_DEPLOYMENT_NAME",
    messages=[{"role": "user", "content": "Hello"}]
)
print(response)

A 403 Forbidden in this case is almost always due to one of these:

• API key ↔ endpoint mismatch
• Wrong deployment name
• Missing IAM role
• Model not deployed / not allowed
• Network restrictions
• Subscription not approved

Please refer this

Azure Portal service health: https://learn.microsoft.com/azure/service-health/overview

Error handling & 403 code details (Python/JS/.NET): – Python: https://learn.microsoft.com/azure/foundry/openai/supported-languages?wt.mc_id=knowledgesearch_inproduct_azure-cxp-community-insider#error-handling

I Hope this helps. Do let me know if you have any further queries.

Thank you!