Unable to deploy sora-2

Hello Taylor Nelson,

Thank you for reaching out and for sharing the details.

I understand you’re having trouble deploying Sora-2 in Azure AI Foundry it looks like you can only see the earlier version (sora, model version 2025-05-02), even though the documentation mentions Sora-2 availability.

Here’s some context and steps to help you move forward:

Why this happens

Even though Sora-2 (model version 2025-10-06) appears in the Azure AI Foundry model catalog, its rollout is still limited access may depend on region, subscription type, or quota. If you only see Sora (v1), that usually means your current region or subscription has not yet been enabled for Sora-2.

1. According to the model catalog, there is indeed a listing for Sora-2: version 2025-10-06 in Azure AI Foundry. Azure AI+2TECHCOMMUNITY.MICROSOFT.COM+2
2. Sora-2’s capabilities are described as including text → video, image → video, video remix/editing, and audio output for synchronized sound. Azure AI+1
3. The original Sora (sometimes listed just as “Sora” or version 2025-05-02) is available for deployment today. Azure AI+1

Here are a few reasons you might not yet see or successfully deploy Sora-2 or use its full capabilities:

The catalog listing for Sora-2 notes that access is restricted (for example enterprise customers or large usage capacity). Azure AI

Newer models often roll out gradually by region, by subscription type (Enterprise vs CSP vs Pay-as-you-go) and by quota. Even though Sora-2 is listed, it may not yet be selectable in your subscription/region.

If you attempted a video generation job expecting audio or remix capabilities (which are part of Sora-2) but you used “model”: “sora” (or your deployment is actually Sora v1), then you’ll see behavior consistent with the “older” model. The fact that audio didn’t work indicates you probably deployed the earlier Sora version.

Sometimes the documentation and catalog show upcoming or preview models, but the production deployment experience may not yet support all features or have universal availability.

Here’s how you can verify your setup and next steps to move forward:

1. Check your deployment model name and version

In the Azure AI Foundry portal (under Models + Endpoints or Model Catalog), verify which model you selected for deployment.

If it shows “sora” or version 2025-05-02, you’re currently using the original Sora (v1).

If Sora-2 is enabled for your region or subscription, you’ll see “sora-2” or version 2025-10-06 listed.

2. Validate feature support

To confirm which model you’re using, try a prompt that involves audio output or video remix features.

If you receive an error like “feature not supported”, it indicates you’re still on the older Sora version.

Sora-2 explicitly supports audio-enhanced video generation and editing as outlined in its documentation.

3. Check region/subscription availability and quota

Quota: Go to Azure OpenAI Studio → Quotas and confirm your standard quota is not exhausted. If needed, request an increase through the Azure portal.

Region: Not all regions currently have access to Sora-2. Check the Azure documentation or reach out to your Azure account team to verify whether your region or subscription tier includes Sora-2 access.

Some customers may need to submit a request form (for example, “Sora-2 access application form”) to gain early access.

4. Verify deployment settings

Make sure your deployment configuration matches Sora-2 requirements:

The model name should be sora-2 (not just sora).

Review any updated API parameters or input schema for the new model in the Azure documentation.

If the model list still doesn’t show Sora-2, try:

Refreshing the Azure AI Foundry portal (Ctrl + F5 or an incognito window) to clear cache.

Checking Azure Monitor logs for any errors like “model not available” or “quota exceeded”, which can help pinpoint the cause.

5. Test via API directly

You can also call the API directly to verify model access:

curl -X POST "https://<your-endpoint>/openai/deployments/sora-2/jobs?api-version=2025-10-06" \
     -H "api-key: <your-key>" \
     -H "Content-Type: application/json" \
     -d '{ "input": { "text": "Generate a short video with ambient audio" } }'

If this returns “model not found”, it confirms your subscription doesn’t yet have Sora-2 enabled.

6. Proceed with Sora v1 and plan for Sora-2 migration

If you need to continue development now, deploy Sora (v1) to keep your workflow moving. Once Sora-2 becomes available in your region, you can:

• Create a new deployment with model = sora-2
• Update the model name and API version in your calls
• Re-test audio/remix functionality before production rollout

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

Thank you!

Is there anywhere that shows current availability? (i.e. which regions) Can I request access via CS Gating? Edit: I found the link to apply for access, however, still, it would be good to show the regions where this can currently be deployed.

Hi Taylor Nelson,

Thank you for following up and for checking the access request link that’s a great step forward.

At this time, Microsoft hasn’t yet published a public list of regions where Sora-2 (model version 2025-10-06) is available. The rollout is still gradual and gated, meaning availability depends on region, subscription type, and capacity.

However, you can confirm your eligibility or request access through the Customer Success (CS) Gating process, which is indeed the correct channel for Sora-2 early access. Once your request is reviewed and approved, the Azure team will enable Sora-2 for your subscription in the supported region.

Thank you1!