MLflow trace storage in Unity Catalog
GAStores MLflow traces in OpenTelemetry-format Delta tables in Unity Catalog, linked to an MLflow experiment, with no per-experiment trace limit, access governed by Unity Catalog table permissions, and traces open to SQL, Genie Code, AI/BI dashboards and any Spark-based tool. OpenTelemetry clients can send spans to a Databricks-hosted OTLP endpoint; since November 2026 Anthropic's Claude Platform can export its traces to Databricks MLflow this way, an Anthropic feature in beta. Since November 2, 2026, a managed warehouse for MLflow (Beta) lets users view traces stored in Unity Catalog in the MLflow UI without choosing a SQL warehouse: MLflow runs the trace queries on a SQL warehouse it creates in the workspace when the workspace supports it, and on serverless compute otherwise; workspace admins turn it on from the Previews page.
Status history
- GA
- Public preview2026-04-30 · Apr 2026Release notes
- Beta2026-01-26 · Jan 2026Release notes
Sources (7)
- 2026-11-05Databricks docs
- 2026-11-03Databricks Blog
- 2026-11-02Release notes
- 2026-10-30Databricks docs
- 2026-07-16Release notes GA
- 2026-04-30Release notes Public preview
- 2026-01-26Release notes Beta
Recent coverage
- 2026-11-04MLflow 3.18.0 release
- 2026-11-03Governing Coding-Agent Sprawl with MLflow AI Gateway news
- 2026-11-03Claude Platform now supports native OpenTelemetry observability in Databricks MLflow news
- 2026-10-27From spot fixes to measured gains: improving document extraction accuracy with ai_extract and MLflow Prompt Registry news
- 2026-10-26How to teach with Databricks Free Edition news
- 2026-10-21MLflow 3.17.2 release
Get Tuesday's version of this
What shipped in Databricks and why it matters, once a week. Free, one-click unsubscribe.
Statuses and dates come from public Databricks sources only. We do not see the private roadmap Databricks shares with customers. If something looks wrong, report a correction.