- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
2 weeks ago
This error is quite specific, and it is probably not a normal permission problem on your side.
The important part is:
Databricks-set rate limit of 0
Databricks is deliberately blocking that model endpoint for your workspace. Databricks documents that a rate limit of 0 effectively disables an endpoint and produces a 403 PERMISSION_DENIED.
Why you're seeing this
Since you said you signed up using AWS, there are two possibilities:
- You created a Databricks Free Edition workspace, or
- You created a Databricks trial workspace through AWS Marketplace.
Both have restrictions around model serving/foundation models. Free Edition has explicit Model Serving limitations, including restrictions on certain models and no GPU serving/provisioned throughput.
The similar problem is been resolved here.
And Databricks documentations also mentions the limitations here.