This assumes that a provider or deployment exists, if you’re adding a new Azure hosted model for the first time, start with Deploying a GPT Model in Microsoft Foundry
| Name | Value |
|---|---|
| Provider adapter name | anthropic_on_aws (for Anthropic models on AWS) |
| Model | The model that was deployed |
| Pretty Name | Name that will show in the provider dropdown on the main page (usually ChatGPT or Claude) |
| Pretty Model | The model that will show in the provider dropdown (usually the same as model) |
| Category | Set the model as Internal (Within your existing data governance) or External (Outside your existing data governance) |
| API Key | AWS Secret Access Key |
| AWS Access Key ID | |
| AWS Access Key ID | |
| Endpoint | The endpoint. Example: |
| us.anthropic.claude-sonnet-4-5-20250929-v1:0 | |
| Tags | Features of that provider |
| disabled → will disable selection in the dropdown | |
| upload → Allow file uploads (discouraged for External models) | |
| dlp → enable dlp (encouraged for External models) | |
| hidden → do not display the selection in the chat dropdown | |
| RAG Search | If tying a provider to a rag model, the rag model would go here |
