Azure Databricks Claude Fable 5 endpoint shows Ready but returns Endpoint not found

Question

Azure Databricks Claude Fable 5 endpoint shows Ready but returns Endpoint not found

Alejandro Morillo 0

In an Azure Databricks workspace in East US 2, the Foundation Model API endpoint databricks-claude-fable-5 is listed as Ready and shows pay-per-token pricing. However, opening it in AI Playground returns: "Endpoint databricks-claude-fable-5 not found". The UI then falls back to another model, so no Fable 5 request is executed.

The model is visible under system.ai. Creating a custom provisioned endpoint fails with: "Model system.ai.databricks-claude-fable-5 is not currently available for serving through this API." Attempting to link deployment metadata fails with HTTP 403 because the registered model is system-owned.

This appears to be a backend entitlement or endpoint-registration synchronization issue: catalog metadata exists, but the pay-per-token serving route is unavailable.

Could a Microsoft/Azure Databricks engineer verify whether Claude Fable 5 is enabled for Azure Databricks in East US 2 and synchronize or enable the Foundation Model API endpoint entitlement/registration for the workspace? I can provide the workspace ID and subscription ID privately if an engineer escalates this to a support case.

SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-07-28T03:45:58.81+00:00
Hi @Alejandro Morillo ,
Based on the behavior described, this appears to be a backend entitlement or endpoint registration synchronization issue. Although databricks-claude-fable-5 is shown as Ready and visible under system.ai, AI Playground returns "Endpoint not found".

Please verify:

Whether Claude Fable 5 is enabled and entitled for the workspace in East US 2.

Whether the Foundation Model API endpoint registration is properly synchronized.

Any recent subscription, entitlement, or endpoint lifecycle changes that could affect model serving availability.

For further investigation, please provide the Workspace ID, Region, timestamp of the failure, and any relevant deployment or serving endpoint logs in private message.
This information will help determine whether the issue is related to model entitlement, availability, or backend registration.
Alejandro Morillo 0 Reputation points

2026-07-28T06:22:01.5766667+00:00

Thank you for confirming the likely backend entitlement/endpoint-registration issue. I have opened a formal Azure support request: 2607280030001390. The case includes the workspace ID, East US 2 region, and permission for Microsoft to collect advanced diagnostic information. Please use that private support case to obtain the workspace identifiers, failure details, and logs, and transfer/escalate it to the Azure Databricks Mosaic AI Model Serving backend team. The failure is reproducible whenever databricks-claude-fable-5 is invoked: the endpoint is shown as Ready, but the request returns "Endpoint databricks-claude-fable-5 not found".
SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-07-29T14:30:54.9266667+00:00
Hi @Alejandro Morillo ,
As backend team suggested Please follow these steps.

An account admin on your side signs in to the account console at https://accounts.azuredatabricks.net, clicks Workspaces, selects your workspace opens the Security and compliance tab, and disables "Enforce data processing within workspace Geography for Designated Services".

Once that setting is changed, retry creating the endpoint or check Playground; Claude Fable 5 should appear.

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.
SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-08-15T09:36:30.8833333+00:00

Hi @Alejandro Morillo ,
I hope you had a chance to review the information shared earlier, and I hope this information has been helpful! If you still have questions, please let us know what is needed in the comments so the question can be answered.

1 answer

Your answer

SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-07-28T03:45:58.81+00:00

Hi @Alejandro Morillo ,
Based on the behavior described, this appears to be a backend entitlement or endpoint registration synchronization issue. Although databricks-claude-fable-5 is shown as Ready and visible under system.ai, AI Playground returns "Endpoint not found".

Please verify:

Whether Claude Fable 5 is enabled and entitled for the workspace in East US 2.

Whether the Foundation Model API endpoint registration is properly synchronized.

Any recent subscription, entitlement, or endpoint lifecycle changes that could affect model serving availability.

For further investigation, please provide the Workspace ID, Region, timestamp of the failure, and any relevant deployment or serving endpoint logs in private message.
This information will help determine whether the issue is related to model entitlement, availability, or backend registration.
Alejandro Morillo 0 Reputation points

2026-07-28T06:22:01.5766667+00:00

Thank you for confirming the likely backend entitlement/endpoint-registration issue. I have opened a formal Azure support request: 2607280030001390. The case includes the workspace ID, East US 2 region, and permission for Microsoft to collect advanced diagnostic information. Please use that private support case to obtain the workspace identifiers, failure details, and logs, and transfer/escalate it to the Azure Databricks Mosaic AI Model Serving backend team. The failure is reproducible whenever databricks-claude-fable-5 is invoked: the endpoint is shown as Ready, but the request returns "Endpoint databricks-claude-fable-5 not found".
SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-07-29T14:30:54.9266667+00:00

Hi @Alejandro Morillo ,
As backend team suggested Please follow these steps.

An account admin on your side signs in to the account console at https://accounts.azuredatabricks.net, clicks Workspaces, selects your workspace opens the Security and compliance tab, and disables "Enforce data processing within workspace Geography for Designated Services".

Once that setting is changed, retry creating the endpoint or check Playground; Claude Fable 5 should appear.

Hope this helps. If you have any follow-up questions, please let me know. I would be happy to help.
SAI JAGADEESH KUDIPUDI 3,640 Reputation points Microsoft External Staff Moderator

2026-08-15T09:36:30.8833333+00:00

Hi @Alejandro Morillo ,
I hope you had a chance to review the information shared earlier, and I hope this information has been helpful! If you still have questions, please let us know what is needed in the comments so the question can be answered.

Answer 1

Hello @Alejandro Morillo ,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that your Azure Databricks Claude Fable 5 endpoint shows Ready but returns Endpoint not found.

Your scenario cannot be solved by recreating or linking the model. The best practice is to check model availability, cross-Geo processing, UC permissions, and the actual serving endpoint route. databricks-claude-fable-5 is a Databricks Foundation Model APIs pay-per-token endpoint, so it should be consumed through the preconfigured model serving route, not by creating or linking a custom/provisioned endpoint. - https://learn.microsoft.com/en-us/azure/databricks/machine-learning/foundation-model-apis/supported-models

What you need to do is to:

Confirm the workspace is in a supported region such as eastus2, where databricks-claude-fable-5 is listed under pay-per-token Foundation Model APIs. - https://learn.microsoft.com/en-us/azure/databricks/machine-learning/model-serving/foundation-model-overview
Enable cross-Geo processing if the workspace is enforcing in-Geo processing, because Claude Fable 5 availability in eastus2 is marked with the cross-Geo indicator. - https://learn.microsoft.com/en-us/azure/databricks/resources/databricks-geos
Verify that the calling user, group, or service principal has EXECUTE permission on system.ai.databricks-claude-fable-5 or on the system.ai schema. - https://learn.microsoft.com/en-us/azure/databricks/machine-learning/foundation-model-apis/model-uc-permissions
Invoke the pay-per-token endpoint directly using the Model Serving invocation API or Anthropic Messages API. [learn.microsoft.com], [docs.databricks.com]
If the endpoint still returns “not found” after those validations, escalate to Azure Databricks support because the remaining cause is backend endpoint registration or entitlement synchronization, not a customer-side deployment issue. - https://learn.microsoft.com/en-us/azure/databricks/machine-learning/model-serving/model-serving-limits, https://docs.databricks.com/api/azure/workspace/servingendpoints/

After cross-Geo processing, Unity Catalog permissions, and the correct pay-per-token invocation route are validated, the endpoint should work. If it does not, then you cannot reliably fix it from the workspace UI or API. Azure Databricks support must verify and synchronize the Foundation Model APIs entitlement and serving route for databricks-claude-fable-5.

I hope this is helpful. Please! Do not hesitate to let me know if you have any other questions, steps or clarifications.

Please do not close the thread by upvoting and accepting the answer if any part of it is helpful.