# JetBrains Releases MPS 2026.2 with ARM64 Support and AI Enhancements The latest MPS update introduces Linux ARM64 compatibility and significant AI agent performance improvements. - **Source:** [The JetBrains Blog](https://brifing.si/source/782eebd1-7be8-49cc-b27c-e552516691b2) - **Published:** 2026-11-10 - **Topic:** [technology](https://brifing.si/c/technology) - **Original article:** - **On Brifing:** https://brifing.si/s/4689a0ba-7f33-490a-859a-9ed496defb69 ## Expanded Platform Support MPS 2026.2 now officially supports Linux on ARM64 architectures, including a bundled JetBrains Java runtime. Developers can create standalone MPS-based applications for this platform using the updated Build Script wizard. This expansion ensures that teams working on ARM64 hardware can leverage the full capabilities of the MPS development environment seamlessly. ## Optimized Archive Packaging The update improves how symbolic links and file permissions are handled during packaging. By adopting .tar.gz for macOS distributions and preserving links instead of duplicating targets, the system reduces redundant file sizes. These changes prevent native libraries from being included multiple times, leading to cleaner and more efficient distribution layouts for developers. ## Resilient Startup Configuration MPS 2026.2 aligns its startup scripts with the IntelliJ Platform by separating boot classpaths from extension classpaths. This modular approach allows for more precise JVM argument management across different operating systems. By reducing manual maintenance, the new launcher configuration minimizes startup errors caused by outdated or overly broad JVM settings during deployment. ## AI Agent Performance Gains The Projectional Agent Toolkit has been overhauled, resulting in over 50% token savings for common tasks like DSL modeling. By implementing a reworked MCP API, the system now uses compact summaries and structure-only metadata instead of full AST dumps. These refinements allow AI agents to operate faster while consuming significantly fewer resources. ## Advanced Type System Tools New MCP tools enable AI agents to interact directly with the type system to debug rules and trace calculations. Agents can now identify the specific rule responsible for errors and determine expected types for roles. This integration provides developers with automated assistance for complex model validation and code authoring tasks. ## Enhanced UI and Navigation Find Usages results now support grouping by virtual package, mirroring the structure of the Project pane. This feature, enabled by default, helps developers locate root nodes in large models more efficiently. Additionally, the Select Up action now recognizes lists of nodes, allowing for more intuitive selection of sibling statements during code editing. ## Refined Testing Capabilities Editor tests now offer flexible result matching, allowing developers to ignore specific node differences or containment roles. Custom matching rules and per-node markers prevent brittle tests from failing due to irrelevant changes. Furthermore, failure messages now explicitly identify the responsible node and cell, eliminating the need for manual debugging of assertions. ## Strategic Development Impact This release focuses on reducing technical debt through better platform integration and more efficient AI tooling. By streamlining the build process and improving agent-model interactions, JetBrains aims to lower maintenance overhead for custom MPS distributions. These updates provide a more stable foundation for teams building complex, language-oriented software solutions today. ## Sources covering this story - https://blog.jetbrains.com/mps/2026/11/mps-2026-2-released/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._