Introducing Apache Spark 4.3 - Now Available in Databricks Runtime 19
Databricks, Wednesday, November 4th, 2026
Databricks announced Apache Spark 4.3 in Databricks Runtime 19, adding stateful processing to Structured Streaming's Real-Time Mode.
Databricks introduced Apache Spark 4.3, available in Databricks Runtime 19, which marks a move to a quarterly feature-release cadence, up from one release every six months.
The headline feature brings stateful processing to Real-Time Mode, reaching sub-150 ms P99 end-to-end latency across three stateful benchmarks, with support for streaming aggregation, dropDuplicates and Scala/Java transformWithState.
Databricks says supported workloads such as fraud detection and device monitoring no longer need a separate streaming engine for millisecond-scale latency.
The release also includes history-preserving Auto CDC and other engine improvements.