An Apache Spark-based analytics platform optimized for Azure.
Hi @Alex Kolchinski ,
Thank you for sharing the detailed troubleshooting results.
Since databricks-gpt-oss-20b can be invoked successfully, the workspace authentication, connectivity, and general Foundation Model API path appear to be functioning correctly. The PERMISSION_DENIED error stating that the endpoint is temporarily disabled due to a Databricks-set rate limit of 0 indicates a Databricks-managed restriction affecting access to those hosted Anthropic models rather than an Azure RBAC, networking, or workspace administrator permission issue.
The READY status confirms the serving endpoint status, but does not necessarily confirm that the workspace is eligible to invoke the model. Similarly, the 404 ENDPOINT_NOT_FOUND response for Claude Fable 5 should be investigated separately, as the error alone is not sufficient to confirm whether the cause is regional availability, workspace eligibility, or another model-specific restriction.
Since workspace-specific entitlements, model access restrictions, and Databricks-managed quotas cannot be verified or changed through the Microsoft Q&A forum, I recommend engaging Databricks Support or your Databricks account team. Please provide the workspace ID, Azure resource ID, endpoint names, timestamps, request IDs, and complete error responses so they can investigate the workspace-specific configuration.
For the billing question, whether these inference charges are eligible to draw from Azure Sponsorship credits depends on the applicable billing configuration and subscription terms. Azure Billing Support and Databricks Support can confirm this for your specific subscription and workspace.
I hope this information helps. If you have any further questions or need additional clarification, please feel free to let me know.