MathWorks R2026b: Simulink Variant Manager, Custom Solvers, and AI-Assisted Profiling Advance Model-Based Design

MathWorks released R2026b on September 24, 2026, delivering a broad set of enhancements to the MATLAB and Simulink product families. The release focuses on three strategic themes: tighter source-control integration for team-based workflows, new tooling for managing product variants in complex system architectures, and expanded AI-assisted capabilities within Simulink Copilot.

New Products in R2026b

Three new products debut in this release:

• Simulink Variant Manager — Provides a dedicated environment for managing product variants across complex system architectures. Engineers working on platform-based designs can now maintain consistency and traceability across variant configurations without manually tracking differences in separate model files.
• Visual Inspection Toolbox — Integrates AI and image-processing techniques to streamline the development and deployment of automated visual inspection applications for industrial and manufacturing settings. The toolbox supports training, validation, and deployment of inspection models directly within the MATLAB environment.
• Polyspace Test Server — Automates unit tests, integration tests, and structural code coverage analysis for C/C++ embedded software within continuous integration (CI) pipelines, enabling teams to enforce quality gates without manual intervention.

Simulink Enhancements

Published Referenced Models

A new "Publish" workflow allows users to create SLXP files—precompiled, ready-to-run versions of referenced models. Published models initialize without requiring rebuilds, which can significantly reduce simulation startup time in large model hierarchies.

Custom Solvers

R2026b introduces the ability to implement custom Simulink solvers. Engineers can now define specialized integration algorithms—such as those based on Chebyshev polynomials or domain-specific numerical methods—and plug them directly into the Simulink simulation engine. This capability is particularly valuable for stiff systems or applications where standard ODE solvers are suboptimal.

JSON Data Dictionaries and TOML Project Files

Simulink Data Dictionaries can now be saved as uncompressed, human-readable JSON files, making diff and merge operations in Git and other version-control systems far more tractable. Complementing this, MATLAB projects can now use a single matlab.toml configuration file instead of the previous multi-file XML approach, reducing repository clutter and simplifying CI configuration.

AI-Assisted Profiling via Simulink Copilot

Simulink Copilot now generates profiling insights to help users identify simulation bottlenecks. Rather than manually interpreting profiler output, engineers can query the Copilot for explanations and optimization suggestions, accelerating the tuning of large or computationally intensive models.

Domain-Specific Highlights

• Aerospace Toolbox and Blockset: Expanded to include high-fidelity orbit propagation, lunar mission support, and enhanced guidance, navigation, and control (GNC) modeling—relevant for both commercial space and defense applications.
• Deep Learning Toolbox: Adds a Time Series Modeler app and enables co-execution of PyTorch models within MATLAB and Simulink, bridging the gap between Python-based ML development and model-based systems engineering.
• Simulink Design Verifier: Now supports distributed analysis across multiple CPU cores, accelerating formal verification for safety-critical embedded software.
• SoC Blockset: Strengthened hardware-software co-design workflows for AMD Adaptive SoCs and FPGAs.
• Point Cloud Toolbox: Extended support for lidar, radar, and SLAM-based perception workflows.

Source Control and Collaboration

Beyond TOML and JSON improvements, the App Designer now saves apps in plain-text formats, and the visdiff function returns a table of changes at the MATLAB Command Prompt—enabling easier integration with coding agents and automated review pipelines. Package management has also been enhanced to support sharing and installing packages directly through File Exchange.

Availability

R2026b is available now through the MathWorks website and via the MATLAB Add-Ons toolbar. Full release notes and migration guides are available at mathworks.com/products/new_products/latest_features.html. The Simulink blog post detailing R2026b Simulink-specific changes is at blogs.mathworks.com/simulink.