Is Apache Spark 4.3 Released?

Not Yet

Apache Spark 4.2 has been released on 14th July 2026

Apache Spark 4.3, the first quarterly feature release, is in the release candidate phase (RC1 on 31st August 2026) and is expected in autumn 2026

Meet Apache Spark 4.2

  • Native GEOMETRY and GEOGRAPHY types
  • Change Data Capture with the CHANGES clause
  • Arrow-optimized Python UDFs by default
  • NEAREST BY top-k join
  • Data Source V2 transaction management
  • SET PATH name resolution
Read more

Read all about Spark

Spark in Action
Stream Processing with Apache Spark
High Performance Spark
Big Data Processing with Apache Spark
Learning Spark
Apache Spark Quick Start Guide

Version

You can check the version of Spark you are using by running the following command in the terminal: 
spark-shell --version

About Apache Spark

Apache Spark is a framework that, like with all of Apache endorsed projects, is open-source, which means that it has an extremely active community of users and developers that need not be direct employees of Spark. Nevertheless, Spark was originally developed by the prestigious University UCBerkeley and was then donated to the Apache foundation. Spark features multiple modules that can be useful to a lot of developers from vastly different backgrounds, mainly in the areas of data science and analysis.

Apache Spark is a unified engine for large-scale data processing: it offers APIs in Scala, Java, Python, R and SQL, so that batch jobs, SQL queries, streaming pipelines and machine learning workflows all run on the same engine. Spark has immense power when developing machine learning applications because of the native module Spark MLlib, which allows complex processes to go smoothly and be at the disposition of the programmer without having to go through the trouble of coding it from scratch or installing and integrating an external module. Spark also has applications in data analysis and visualizations with the native modules that are installed in the framework.

Some of the main features of Spark are crucial to understanding. Spark is modular, meaning that you only need the components you require (Spark SQL, Structured Streaming, MLlib or GraphX) on top of the core engine, and it keeps data in memory across the cluster, which is what makes it so much faster than classic MapReduce jobs on Hadoop.

Releases

Spark was launched as we know it on the 30th of May of 2014, but there are some early development releases like version 0.5, which was made available to the public as early as 2012. In fact, the origins of Spark trace back to development in UCBerkeley as early as 2009, where the idea began forming for this tool. After quite some time and a lot of developer efforts, Spark is currently on the 4.2.0 release, made available on 14th July 2026; Spark 4.3.0, the first of the new quarterly feature releases, is in its release-candidate phase. Spark has seen so much popularity that an online teaching and course community is forming around the tool, pointing to future widespread usage and more updates to be expected from the developer team.