Artifacts

Artifacts are the inputs and outputs of pipeline steps. When a pipeline run completes, each step can produce artifacts — trained models, evaluation metrics, data files, or any other output. Artifacts are tracked automatically and linked to the run and step that created them.

Preview

The Pipelines tab is in early access. Artifact browsing is read-only in the UI.

The artifacts list

The list view shows every artifact across all pipeline runs. Each row displays:

• Name — the artifact's display name.
• Type — artifact type (model, dataset, metric, etc.).
• Pipeline — which pipeline produced it.
• Run — which run produced it.
• Created — when the artifact was created.

Click an artifact to see its metadata, lineage, and download link.

Next steps

• Runs — the executions that produce artifacts
• Executions — low-level step executions