← Back
Databricks
Databricks ships February release with runtime maintenance, Python testing, and new models
· releasefeatureapibugfixsecurityperformancemodelplatformintegration · docs.databricks.com ↗

Runtime Maintenance and Performance

Databricks Runtime maintenance updates are now available across all supported versions, including Runtime 18.0, 17.3 LTS, 17.2, 16.4 LTS, 15.4 LTS, 14.3 LTS, 13.3 LTS, and 12.2 LTS. These updates include bug fixes, security patches, and performance improvements. Customers should review the release notes for each runtime version to identify applicable patches for their deployments.

Developer Experience Improvements

Python unit testing in the workspace is now generally available, providing integrated testing tools directly in the Databricks IDE. Developers can discover, run, and debug pytest-based tests using the tests sidebar, inline execution glyphs, and a dedicated testing results panel. This streamlines the development workflow for Python-based data and ML projects.

Model Availability Expansion

Foundation Model APIs now support OpenAI GPT-5.3 Codex and Google Gemini 3.1 Pro as Databricks-hosted models. GPT-5.3 Codex specializes in complex code generation, large-scale refactoring, and enterprise software engineering tasks. Both models are available through Foundation Model APIs on a pay-per-token basis.

Infrastructure and Compliance

New compliance controls include ISMAP compliance support for Japanese government certification requirements and HITRUST compliance controls (currently in Public Preview) for risk management and security/privacy demonstration. Additionally, serverless outbound IPs are now available through a public JSON endpoint (Public Preview), replacing the previous stable IP approach for firewall configuration.

API and Tooling Updates

JAR tasks on serverless compute entered Public Preview, enabling Java-based jobs to run on serverless infrastructure. The ADBC driver is now the default for new Power BI connections, replacing ODBC with Arrow Database Connectivity for improved performance. Existing connections remain on ODBC unless manually updated. The Databricks ODBC Driver was renamed from Simba Spark ODBC Driver; existing Simba versions remain supported for two years.

SQL and Security Enhancements

COMMENT ON TABLE for streaming tables is now generally available, allowing developers to add or update documentation directly on table objects. Personal access token behavior was updated so that revoking CAN USE permissions deactivates (but doesn't delete) tokens—restoring the permission reactivates the same tokens.