An Apache Spark-based analytics platform optimized for Azure.
Hiello @Florimed
Thank you for reaching out to Microsoft Q&A.
Based on the behavior shown, this appears to be related to model availability/tenant enablement rather than an issue with the Databricks workspace configuration itself.
From your screenshots: the older tenant/workspace shows Anthropic models such as Claude Opus, Sonnet, and Haiku, while the newer tenant currently only shows OSS/open models such as GPT OSS, Llama, and Gemma.
This usually indicates that Anthropic Foundation Model APIs (pay-per-token) are not yet enabled or exposed for the newer tenant/workspace.
A few important things to verify:
- Regional support for Foundation Model APIs (pay-per-token)
Anthropic models are not available in all Azure Databricks regions yet.
Please verify whether: West Europe currently supports Anthropic Foundation Model APIs (pay-per-token), or whether support is still limited to selected regions.
If West Europe is not fully enabled yet, possible options are deploying a workspace in another supported region, or contacting Databricks support/account team to confirm rollout status for West Europe.
Workspace “Data Processing” restrictions
If the workspace has: “Enforce data processing within workspace geography” enabled, some system-provided foundation models including Anthropic models may not appear in the endpoint catalog.
If enabled, temporarily disabling the restriction and refreshing the endpoint list may allow the Claude models to appear.
Marketplace / entitlement validation
Please also confirm: required Marketplace terms for Anthropic models have been accepted, the subscription does not have policy/security restrictions, and the tenant has appropriate entitlement for third-party foundation models.
Compare both tenants/workspaces
Since the older tenant already works, it would help to compare: workspace region, subscription type, compliance policies, preview feature enrollment, and admin data-processing settings.
At this time there is no direct UI toggle to manually enable Claude models if the tenant has not yet been onboarded for them.
Please refer this
Model serving features availability (regions + feature flags) https://learn.microsoft.com/azure/databricks/resources/feature-region-support#azure-model-serving
Cross-Geo data processing restrictions https://learn.microsoft.com/azure/databricks/resources/databricks-geos#cross-geo-processing
Foundation Model APIs supported models https://learn.microsoft.com/azure/databricks/machine-learning/foundation-model-apis/supported-models
Model serving endpoint permissions https://learn.microsoft.com/azure/databricks/machine-learning/model-serving/manage-serving-endpoints#manage-permissions-on-a-model-serving-endpoint
Monitoring Azure Databricks release notes https://learn.microsoft.com/azure/databricks/release-notes/product/
I Hope this helps. Do let me know if you have any further queries.
Thank you!