Model Repository API

The Model Repository API manages catalog-backed AI and ML model records and their model artifacts. It is the automation surface for the Model Repository user experience: listing space or organization models, creating and updating model metadata, uploading large artifacts in chunks, verifying checksums, downloading artifacts, and deleting artifacts.

info

All endpoints are mounted under /api/v1.0. Paths below omit that prefix for readability.

Permissions

Model records are catalog products. The required access level depends on the operation:

Operation	Required access
List space models	Access to the space.
List organization models	Access to the organization; results include spaces the caller can access.
Create model	Access to the target space and available catalog product capacity.
Update model metadata	edit access on the product.
List or download artifacts	view access on the product.
Upload, complete, or delete artifacts	edit access on the product.

Endpoint Overview

Method	Endpoint	Description
GET	/catalog/spaces/{slug}/model-repository/models	List model repository records in a space.
GET	/organizations/{slug}/model-repository/models	List model repository records across accessible organization spaces.
POST	/catalog/spaces/{slug}/model-repository/models	Create a model repository record in a space.
PATCH	/catalog/products/{product_id}/model-repository	Update model repository metadata for a product.
GET	/catalog/products/{product_id}/model-artifacts	List model artifacts attached to a product.
POST	/catalog/products/{product_id}/model-artifacts/uploads	Initiate a managed artifact upload session.
PUT	/catalog/products/{product_id}/model-artifacts/uploads/{upload_id}/chunks/{chunk_index}	Upload one artifact chunk.
POST	/catalog/products/{product_id}/model-artifacts/uploads/{upload_id}/complete	Complete an upload session and create the artifact.
GET	/catalog/products/{product_id}/model-artifacts/{artifact_id}/download	Download an artifact or follow a storage redirect.
DELETE	/catalog/products/{product_id}/model-artifacts/{artifact_id}	Delete an artifact.

Model Filters

Space and organization listing endpoints share the same query parameters.

Parameter	Type	Description
skip	integer	Number of records to skip. Default: 0.
limit	integer	Number of records to return. Default: 50, maximum: 1000.
q	string	Search by model text fields.
model_family	string	Filter by model family.
types	array	Repeatable product type filter, for example types=ML%20Model&types=AI%20System.
deployment_status	string	Filter by deployment status.
model_task_type	string	Filter by task type.
risk_classification	string	Filter by risk classification.
include_archived	boolean	Include archived products. Default: false.

List responses include pagination headers:

Header	Description
X-Total-Count	Total matching records.
X-Page	Current page number derived from skip and limit.
X-Has-More	Whether more matching records exist after this page.

Supported product types are ML Model, AI System, LLM Agent, AI App, Agent, and Prompt Collection.

Supported model families are traditional_ml, genai, foundation_model, embedding_model, agent, ai_app, prompt_collection, and other.

Common task types include llm, classification, regression, clustering, forecasting, embedding, ranking, anomaly_detection, recommendation, agent, and other.

List Space Models

GET /catalog/spaces/{slug}/model-repository/models

Returns model repository summaries for a single space.

curl -X GET "https://api.qarion.com/api/v1.0/catalog/spaces/ml-platform/model-repository/models?model_family=genai&limit=50" \
  -H "Authorization: Bearer YOUR_API_KEY"

List Organization Models

GET /organizations/{slug}/model-repository/models

Returns model repository summaries across the organization spaces the caller can access.

curl -X GET "https://api.qarion.com/api/v1.0/organizations/acme/model-repository/models?types=ML%20Model&include_archived=false" \
  -H "Authorization: Bearer YOUR_API_KEY"

Create Model Metadata

POST /catalog/spaces/{slug}/model-repository/models

Creates a catalog product using a Model Repository-supported product type and returns a ModelRepositoryModelSummary.

{
  "name": "Churn Prediction Model",
  "slug": "churn-prediction-model",
  "description": "Predicts account churn risk for customer success workflows.",
  "product_type": "ML Model",
  "provider": "ml-platform",
  "environment": "production",
  "risk_classification": "limited",
  "lifecycle_stage": "monitoring",
  "model_family": "traditional_ml",
  "model_task_type": "classification",
  "model_architecture": "XGBoost",
  "framework": "scikit-learn",
  "intended_use": "Prioritize customer success outreach.",
  "limitations": "Not calibrated for accounts with less than 30 days of activity.",
  "deployment_status": "production",
  "serving_url": "https://models.example.com/churn",
  "input_schema": {
    "type": "object",
    "required": ["account_id", "usage_30d"]
  },
  "output_schema": {
    "type": "object",
    "properties": {
      "churn_probability": {
        "type": "number"
      }
    }
  }
}

Create Fields

Field	Type	Required	Description
name	string	Yes	Display name for the model product.
slug	string	No	URL-friendly product slug.
description	string	No	Short repository and catalog summary.
long_description	string	No	Extended model description.
product_type	string	No	Defaults to ML Model. Must be a supported model product type.
provider	string	No	Source platform or provider.
environment	string	No	Defaults to development.
risk_classification	string	No	Governance risk classification.
lifecycle_stage	string	No	Lifecycle stage such as design, training, validation, deployment, monitoring, or retirement.
model_family	string	No	Model family filter value.
model_task_type	string	No	Model task type.
model_architecture	string	No	Architecture or algorithm family.
framework	string	No	Training or serving framework.
intended_use	string	No	Intended business or technical use.
limitations	string	No	Known limits, failure modes, or exclusions.
deployment_status	string	No	Defaults to development.
serving_url	string	No	Serving endpoint or operational URL.
input_schema	object	No	Input schema metadata.
output_schema	object	No	Output schema metadata.
ai_system_type	string	No	AI system classification where applicable.
vendor_name	string	No	External vendor name.
vendor_product_name	string	No	External vendor product name.
vendor_model_id	string	No	External vendor model identifier.
deployment_environment	string	No	Deployment environment label for vendor or platform deployments.
business_use_case	string	No	Business use case for governance context.
is_gpai	boolean	No	General-purpose AI indicator.

Update Model Metadata

PATCH /catalog/products/{product_id}/model-repository

Updates the Model Repository metadata for an existing model product. The request body accepts the same fields as create, all optional. Include only fields to change.

{
  "deployment_status": "retired",
  "lifecycle_stage": "retirement",
  "limitations": "Retired after the v3 model replaced this workflow."
}

Model Summary Response

List, create, and update operations return ModelRepositoryModelSummary objects.

Field group	Fields
Identity	id, slug, name, description, product_type, provider, space_id, space_slug, space_name
Governance and lifecycle	risk_classification, lifecycle_stage, environment, deployment_status
Model metadata	model_family, model_task_type, model_architecture, framework, serving_url, input_schema, output_schema, ai_system_type, vendor_name, vendor_product_name, vendor_model_id
Version summary	version_count, current_version_id, current_version_number, current_version_type, current_version_release_date
Artifact summary	artifact_count, current_artifact_id, current_artifact_file_name, current_artifact_kind, current_artifact_content_type, current_artifact_model_format, current_artifact_sha256, current_artifact_size_bytes
Timestamps	created_at, updated_at

List Artifacts

GET /catalog/products/{product_id}/model-artifacts

Returns all non-deleted model artifacts attached to the product.

curl -X GET "https://api.qarion.com/api/v1.0/catalog/products/550e8400-e29b-41d4-a716-446655440000/model-artifacts" \
  -H "Authorization: Bearer YOUR_API_KEY"

Upload Artifacts

Managed artifact uploads use a three-step flow.

1. Initiate Upload

POST /catalog/products/{product_id}/model-artifacts/uploads

{
  "file_name": "churn-model-v2.onnx",
  "content_type": "application/octet-stream",
  "total_size_bytes": 73400320,
  "sha256": "0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
  "artifact_kind": "model_binary",
  "version_label": "v2.0.0",
  "product_version_id": "4c1e92ac-e430-4b4e-ae49-3a1111111111",
  "training_run_id": "5c1e92ac-e430-4b4e-ae49-3a2222222222",
  "is_current": true,
  "metadata": {
    "format": "onnx",
    "python": "3.11"
  }
}

Field	Type	Required	Description
file_name	string	Yes	Original file name.
content_type	string	Yes	MIME content type.
total_size_bytes	integer	Yes	Total file size in bytes.
sha256	string	No	Expected SHA-256 hex digest.
artifact_kind	string	No	Defaults to model_binary.
version_label	string	No	Human-readable artifact version label.
product_version_id	UUID	No	Product version to associate with the artifact.
training_run_id	UUID	No	Training run to associate with the artifact.
is_current	boolean	No	Mark the completed artifact as current. Default: false.
external_uri	string	No	External artifact URI metadata.
metadata	object	No	Additional artifact metadata.

The response is a ModelArtifactUploadSessionResponse.

Field	Description
id	Upload session ID.
product_id	Product receiving the artifact.
file_name	File name from the initiation request.
content_type	MIME content type.
total_size_bytes	Expected total upload size.
chunk_size_bytes	Chunk size the client should use.
expected_sha256	Expected SHA-256 digest, when supplied.
received_bytes	Bytes received so far.
received_chunks	Chunk indexes received so far.
status	initialized, uploading, completed, failed, or expired.
expires_at	Expiration timestamp for the upload session.
created_at	Session creation timestamp.
completed_at	Completion timestamp, if completed.

2. Upload Chunks

PUT /catalog/products/{product_id}/model-artifacts/uploads/{upload_id}/chunks/{chunk_index}
Content-Type: application/octet-stream

Send the raw bytes for the chunk in the request body. Use zero-based chunk indexes and the chunk_size_bytes returned by the initiation response.

curl -X PUT "https://api.qarion.com/api/v1.0/catalog/products/550e8400-e29b-41d4-a716-446655440000/model-artifacts/uploads/660e8400-e29b-41d4-a716-446655440000/chunks/0" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/octet-stream" \
  --data-binary "@chunk-0.bin"

Each successful chunk upload returns the updated upload session.

3. Complete Upload

POST /catalog/products/{product_id}/model-artifacts/uploads/{upload_id}/complete

{
  "sha256": "0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
  "is_current": true
}

Completion assembles the uploaded chunks, computes SHA-256, validates the digest when one was provided, stores the artifact, and returns a ModelArtifactResponse.

Artifact Kinds

Value	Description
model_binary	Serialized model binary or package.
weights	Model weights.
checkpoint	Training or fine-tuning checkpoint.
adapter	Adapter, delta, or fine-tuning artifact.
tokenizer	Tokenizer or vocabulary artifact.
prompt_bundle	Prompt collection or prompt bundle artifact.
eval_report	Evaluation report or benchmark artifact.
model_card	Model card or release documentation.
other	Other model artifact.

Artifact Response

Artifact endpoints return ModelArtifactResponse objects.

Field	Description
id	Artifact ID.
product_id	Product that owns the artifact.
product_version_id	Linked product version, if any.
training_run_id	Linked training run, if any.
artifact_kind	Artifact kind.
version_label	Human-readable artifact version label.
file_name	Artifact file name.
content_type	MIME content type.
size_bytes	Artifact size in bytes.
sha256	Computed SHA-256 digest.
storage_backend	Storage backend that holds the artifact.
storage_uri	Backend storage URI.
external_uri	External artifact URI metadata, if supplied.
status	available, deleted, or failed.
is_current	Whether this artifact is the current artifact for the model.
metadata	Additional artifact metadata.
uploaded_by_id	User who uploaded the artifact.
uploaded_at	Upload timestamp.
created_at	Record creation timestamp.
updated_at	Last update timestamp.

Download Artifact

GET /catalog/products/{product_id}/model-artifacts/{artifact_id}/download

Returns the artifact file when it is stored locally, or a redirect response when the configured storage backend provides a download URL.

Delete Artifact

DELETE /catalog/products/{product_id}/model-artifacts/{artifact_id}

Deletes the artifact record and returns 204 No Content.

Error Notes

Condition	Typical response
Caller lacks the required access.	403 Forbidden.
Organization, product, upload session, or artifact cannot be found or accessed.	404 Not Found or a service-specific error response.
Request payload fails validation.	Validation error response from the API.
Space has reached its catalog product limit.	Limit enforcement error from the catalog tier gate.
Uploaded bytes do not match the expected SHA-256 digest.	Upload completion fails with a validation error.
Artifact file is missing from storage.	404 Not Found with Artifact file not found.

Related Docs

• Model Repository user guide
• Data Catalog API
• Product Types API