A catalog of AI models in Microsoft Foundry that you can discover, compare, and deploy using Azure’s built‑in tools for evaluation, fine‑tuning, and inference
Hello Hongwei Yu,
Thanks for reaching out to Microsoft Q&A,
It looks like you’ve successfully added claude-sonnet-4-6 / 4-5 in Azure AI Foundry, and the portal shows Succeeded, but when calling from VS Code you get:
“The model claude-sonnet-4-6 is not available on your Foundry deployment.”
This typically means the issue is not deployment itself, but a mismatch between:
- Deployment vs API usage
- Environment (region/subscription)
- Or client configuration (VS Code)
There are four main categories where things break in this scenario:
1) Deployment Name vs Model Name
Even though you see: Model - claude-sonnet-4-6
APIs do not use the model name
They use the deployment name (the “Name” column in your screenshot).
What to check
Go to Foundry → Models → Deployments
Copy the exact deployment name, then use:
"model": "<your-deployment-name>"
NOT:
"model": "claude-sonnet-4-6"
- Subscription & Region Constraints
Claude models in Foundry are not globally available.
Requirements:
Paid Azure subscription
Supported regions:
- East US 2
- Sweden Central
If your resource is in another region deployment may show “Succeeded” but won’t be callable
3) Deployment Type & Readiness
There are two ways models appear in Foundry:
- “Added” in catalog
- “Deployed” + READY
Please verify
- Go to Models → Models from partners
- Select Claude Sonnet 4.6 (Global standard)
- Click Deploy
- Wait until status - READY
“Succeeded” alone isn’t always enough ensure it’s actually serving
4) Endpoint Configuration
When using Claude via Foundry, your endpoint must include Anthropic routing:
https://<your-resource>.services.ai.azure.com/anthropic
If your VS Code config is pointing to:
- OpenAI endpoint
- Anthropic direct API
It won’t find your deployment
5) Permissions / Access Control
Even if deployment is correct, access can fail silently.
Ensure Role assigned Contributor OR Owner on resource
If using Entra ID Assign Cognitive Services User role via Microsoft Entra ID
- SDK / VS Code Compatibility
Some VS Code extensions:
- Expect OpenAI-style APIs
- Don’t fully support multi-provider routing in Foundry
Model exists - but tool says “not available”
How to confirm
Test outside VS Code:
- Python SDK (
azure-ai-projects) - Postman / REST
If it works there issue is with VS Code extension, not Foundry
- Validate via SDK
models = client.models.list()
If your deployment is not listed it’s not exposed to your runtime context
- Check Deployment Logs
Go to Foundry → Jobs → Deployment Jobs
Look for Region mismatch, Permission errors, Provisioning failures
Please refer this
Subscription type & region support for Claude: https://learn.microsoft.com/azure/foundry/foundry-models/how-to/use-foundry-models-claude#subscription-type-and-region-support
Common troubleshooting for Claude in Foundry: https://learn.microsoft.com/azure/foundry/foundry-models/how-to/use-foundry-models-claude#troubleshooting
I Hope this helps. Do let me know if you have any further queries.
Thank you!