Get Model Lineage

Retrieve the full lineage graph for a model, including training datasets and parent models.

Returns the complete lineage graph for a model attestation. The graph includes all training datasets, parent models (for fine-tuned or distilled models), and downstream deployments connected by typed relationships.

Model lineage enables compliance teams to trace exactly which data was used to train a model, verify that all upstream datasets are properly attested, and audit the full provenance chain from raw data to production deployment.

Authentication

X-API-Key · string · required

API key with models:read scope. Alternatively, pass a Bearer JWT token in the Authorization header.

X-Tenant-ID · string · required

Tenant identifier for multi-tenant isolation.

Path Parameters

id · string · required

Model attestation identifier (maip-model:ULID).

Response

model_id · string

The root model identifier for this lineage query.

nodes · array

Array of lineage nodes. Each node contains: - id (string) — Resource identifier - type (string) — Node type: dataset, model, or deployment

• name (string) — Human-readable name - version (string) — Resource version - status (string) — Current attestation status - created_at (string) — ISO 8601 timestamp

edges · array

Array of relationships connecting nodes. Each edge contains: - source (string) — Source node identifier - target (string) — Target node identifier - relationship (string) — Relationship type (e.g. trained_on, fine_tuned_from, distilled_from, deployed_as) - metadata (object) — Optional edge metadata

training_datasets · array

Convenience field listing all dataset nodes in the lineage that have a direct trained_on relationship with the queried model. Each entry contains id, name, version, and status.