List Aligned Models

Retrieve all domain-aligned models for the authenticated user.

This endpoint returns a list of custom models that have been created through domain alignment workflows, including their deployment status, evaluation results, and conversation settings.

Returns:

• domain_aligned_models: List of aligned model objects, each containing:
  • id: Unique model identifier (or alignment-{id} for un-deployed models)
  • alignment_id: Original alignment project ID
  • name: Model name
  • base_model: Base model used for alignment
  • alignment_project: Name of the alignment project
  • created_date: Creation timestamp
  • status: Model status (e.g., DEPLOYED, UNDEPLOYED, DEPLOYING, UNDEPLOYING, EVALUATED)
  • deployment_status: Current deployment state (DEPLOYED or UNDEPLOYED)
  • endpoint: Inference endpoint URL (only for deployed models)
  • creator: Username/email of the model creator
  • usage_count: Number of times the model has been used for inference
  • last_used: Timestamp of last usage
  • performance_metrics: Model performance statistics
  • downloadable: Whether model can be downloaded
  • evaluation_data: Benchmark evaluation results (if available)
  • conversation_saving_enabled: Whether conversation saving is enabled for this model

Example Request:

GET /api/v3/models/aligned
Headers: {"Authorization": "Bearer <api_key>"}

Example Response:

{
  "domain_aligned_models":
    [
    {
      "id": "aligned-model-01kmqm4nrn9fw6r",
      "alignment_id": "alignment-xyz789",
      "name": "customer_support_model",
      "base_model": "nugen-flash-instruct",
      "alignment_project": "Customer Support Alignment",
      "created_date": "2024-01-15T10:30:00Z",
      "status": "DEPLOYED",
      "deployment_status": "DEPLOYED",
      "endpoint": "https://api.example.com/v1/models/alignment-xyz789",
      "creator": "user@example.com",
      "usage_count": 150,
      "last_used": "2024-01-20T14:30:00Z",
      "performance_metrics": {
        "accuracy_before": 0.75,
        "accuracy_after": 0.92,
        "uncertainty_before": 0.35,
        "uncertainty_after": 0.15,
        "domain_violations_before": 12,
        "domain_violations_after": 2
      },
      "downloadable": true,
      "evaluation_data": {
        "accuracy": 0.92,
        "total_questions": 50
      },
      "conversation_saving_enabled": true
    }
    ]
}

Notes:

• Returns both deployed and undeployed aligned models
• evaluation_data is included if the model was evaluated against a benchmark
• performance_metrics show before/after alignment improvements
• conversation_saving_enabled indicates whether chat history is saved for each model
• Use /models/base to see available base models for alignment