Introducing STM32CubeMX2, a new flavor of STM32CubeMX tool

[Maxime_MARCHETTO1]

Introduction

Building on the solid foundation of STM32CubeMX, STM32CubeMX2 takes the STM32 development experience to the next level. It enhances proven functionalities and introduces new features to deliver greater efficiency, flexibility, and innovation from its very first release.

• Coexistence with STM32CubeMX: STM32CubeMX and STM32CubeMX2 will coexist. STM32CubeMX2 is introduced for new STM32 series, starting with STM32C5.
• Future ready workflows: STM32CubeMX2 continues to streamline existing workflows while opening the door to more advanced capabilities.
• Active evolution: ST will keep enhancing STM32CubeMX2 with regular fixes and new features over time.

live code preview for faster prototyping

The new live code preview in STM32CubeMX2 accelerates prototyping for developers who use STM32CubeMX as a code snippet generator:

• No need to generate full projects
  Get the initialization code directly in the STM32CubeMX2 live code preview window, ready to paste into your existing project.
• Instant feedback
  Any change in the configurator is immediately reflected in the code preview, so you can track updates in real time.
• Diff view for clearer understanding
  A built-in diff function shows how configuration changes impact the generated code, helping you understand updates at a glance.
  
  

Highlight Path: clearer clock configuration

Tracing clock signals with your finger on the screen is no longer necessary. In STM32CubeMX2, the Highlight Path feature shows:

• Which peripherals depend on a given clock
• Which clock a specific peripheral is running on

In addition, a new table view provides a simple, organized display of all clocks, making it much easier to understand and manage your clock configuration.

Pinout table view for faster, clearer configuration

On the configuration side, the new pinout experience boosts productivity with:

• Enhanced pin search
• Clear highlighting of alternative pins for a given I/O
• A table view that makes assignments easier to review and adjust

These improvements help shorten time to market while keeping project dependencies and configuration under control. The optimized graphical interface makes it easier to identify the right pin for your design, enabling hardware and software teams to collaborate more efficiently and get projects ready faster.

GUI updates for a smoother experience

The new and enhanced user interface is designed to make your workflow more intuitive and efficient. By prioritizing easy access to the most frequently used and straightforward settings, it streamlines the journey from concept or datasheet to a working prototype—so you can focus on what matters most.

With a new quick menu, you’ll never miss an important setting. It ensures you can always find what you’re looking for, even when navigating more complex features, making your workflow smoother and more efficient.

STM32CubeMX2 documentation goes online

Instead of relying on static PDF files, STM32CubeMX2 now features a brand-new online documentation portal, ensuring you always have access to the latest, most up-to-date information.

The portal also provides direct links to articles, application notes, and related resources, helping you achieve more in less time.

New pack manager to keep projects up to date

STM32CubeMX2 introduces packs to automatically manage dependencies when building a project. This makes it easier to:

• Keep your projects aligned with the latest software versions
• Add middleware or hardware components as needed
• Generate code based on LL or HAL2, the new hardware abstraction layer deployed with new STM32 series starting from STM32C5

STM32CubeMX2 also supports multiple project output formats to match your toolchain strategy, including:

• STM32CubeIDE
• CMake-based workflows (such as VS Code and other IDE/compiler environments)
• Partner IDE formats, including IAR and Keil Studio
  
  

Known limitations

This first STM32CubeMX2 release is a major milestone in our roadmap. We are aware of some current limitations, but we wanted to deliver these new capabilities to you as early as possible. The tool is fully functional, and its output can be safely used for production software. Regular patches and feature updates are already planned to improve the experience.

The most visible gaps today relate to performance with large configurations, some IDE/toolchain and regeneration workflows that still require manual steps, and the fact that it is not yet possible to lock a specific component version.

Performance

• Large projects or MCUs may feel slower than expected.
  Ongoing optimizations are targeting improved responsiveness and scalability.
  
  

Packs and middleware

• Pack import and compatibility visibility are still improving; some legacy packs may not be configurable.
• Currently, FreeRTOS is the only supported middleware. Other middleware components will be added over time.
  
  

IDEs, toolchains, and regeneration

• STM32CubeMX2 cannot track changes made manually to project files outside the tool.
• Manual edits done in the IDE can be overwritten when regenerating the project—this is particularly true for IAR, where the entire project is contained in a single file that STM32CubeMX2 replaces.
• To mitigate this risk, we strongly recommend:
• Using a version control system
• Enabling the available settings to back up user files during project generation
  
  

Configuration and user experience

• Some workflows are still more complex than desired.
• Certain conveniences—such as centralized privilege management or flexible init ordering—are not yet available.
  
  

Project generation and files

• Some generated artifacts follow fixed conventions (for example, file locations).
• Certain defaults (such as GPIO speed) may need manual adjustment depending on peripheral requirements.
  
  

Hardware parts / start from board

• Hardware-part coverage and board-default activation are currently more limited than our long-term target.
  
  

Installation and updates

• STM32CubeMX2 does not self-update.
• Upgrades can require a reinstall, and parallel installations are not supported.

We plan to deliver quarterly releases to expand device and middleware support, improve UX and performance, and progressively close these known gaps.

For the most up to date list of known limitations, please refer to the online documentation portal.

Additional resources

• Get started with STM32C5 MCU
• Download STM32CubeMX
• Video: Introduction to the STM32Cube ecosystem
• Video: [Getting started] STM32CubeMX

First published on Mar 16, 2026

[Niclas Hedhman]

This Stm32CubeMX2 release seems to be a rather quite announcement, but STM32World has already made an unbiased review of this latest train wreck from ST. https://www.youtube.com/watch?v=61r83TTzDfk

[Suess]

Super cool! This version is what I've been looking for where it doesn't fully regenerate the project files, and it keeps you inside of VS Code. Really hope you folks reconsider and include previous models such as the STM32H5. Still, can't wait to test drive this feature in the meantime.

The only hiccup we run into with "MX1" is having to rolling back file formatting for +87 files that either conflict with line endings (CRLF / LF / CR), tab style, and then reformat the 2 files with clang-format that were actually modified.

Looking forward to seeing the new stuff come down the pipeline. Keep up the great work!

[Grufs]

Since we're not using the STM32C5, this announcement is still mostly of academic interest to us.

However, I just cannot understand why I should have to download gigabytes of STM32cubeMX Java Bytecode, just to have it configure a starting point for my microcontroller coding.

I have little problem with my computer executing Bytecode, but I do not appreciate it – and its runtime interpreter – taking up precious space on my SSD. [Actually, it's been relegated to a separate, hand-me-down computer.]

Why is the STM32CubeMX not a Web [assembly] application, hosted on ST servers? It's not like anyone will be lingering for long in this sort of application, taking up excessive amounts of server capacity. 

[MRich.2]

@Grufs, interesting that you like the idea of making STM32CubeMX a web application.  That sounds awful to me, but I'd like to know more about why you think that is a great idea.  For me, my 1-2 TB SSDs have plenty of room for applications.  Keeping my work local seems better from a cyber security point of view.  I worry if configurator was in the cloud it would slow down my workflow waiting to push and pull wads of data between my computer and the cloud.  It seems like adding another step to a process that already seems to be getting larger and less convenient rather than smaller and more convenient.  Generally, I find web applications require a lot more mouse clicking and are a lot slower than my desktop applications + keyboard shortcuts.  I need to be able to pick and choose the version of MX I'm using and I need to be able to pick and choose the version of the processor library I'm using - presumably a cloud application would still be able to support this aspect though.  And for product lifecycle maintenance, we archive tool versions that our product depends on so that we can continue to maintain them into the future - even if the tool company goes out of business (the idea is to buy time in case of emergencies).  I don't see how that would be possible with a web tool which I suppose would just go away if the company went out of business or decided to just stop supporting it.    

[mrkozmic]

Will this be available in CubeIDE as well?

[Etienne_Parmentier]

Will it ever be easier to set up a multi-board project ? We have 2 dev board and using cmake to build 2 configurations (of the same CPU with different packaging) and it's not easy at all to setup. You essencially have to maintain 2 branches or use git submodules which is even more painfull.

Of course switching between 2 MCU is much harder and non trivial, but being able to switch between compiling for a nucleo MCU and the one you actually use in your pcb design could be better !

[DClif.1]

Feedback:
1. Having two programs is very inconvenient

a) when using the MCU selector to find suitable devices.

b) we all now how to learn how to use two programs instead of one.

Request that you consolidate the two programs and support all existing MCUs that CubeMX supports in CubeMX2.

 

2. Code is not important at all when using CubeMX to design pin-layouts, check for pin-compatible devices, and so on.  Code is even less important when using a either a) a different hardware library like OpenCM3, b) when using a more modern language, like RUST for software development.

 

[Ahmet Yasin CİVAN]

Great update, but we can do more in the AI era!

First of all, the announcement of STM32CubeMX2 is great news. Features like the "Live Code Preview" and the "Highlight Path" in clock configurations will definitely speed up the prototyping process and enhance the developer experience. Congratulations to the whole team!

However, when talking about a completely new foundation and a "future-ready" workflow, I think there is a huge potential missing from the table: Native AI (LLM) support directly integrated into STM32CubeMX2. We are living right in the middle of the AI era, and hardware development tools need to keep pace. We know ST has stepped into this area recently with the STM32 Sidekick tool. But integrating such powerful LLM capabilities directly into CubeMX2's core configuration workflow would make things incredibly easy.

Here are a few LLM / AI-powered feature suggestions that could be added to STM32CubeMX2 for future roadmap updates:

1. Natural Language Hardware Configuration (Prompt-to-Config): Instead of spending time looking for the right menu or pin in the UI, we should just be able to type to an integrated assistant: "Configure I2C1 as 400kHz Fast Mode on pins PB6 and PB7" or "Optimize the system for the lowest power consumption for STM32C5". The LLM should understand this command and instantly apply the configuration to the GUI.

2. Smart Conflict Resolution: When there is a conflict in pinout or clock configurations, the assistant could analyze the root cause in seconds and propose a direct solution: "The X pin you are using is conflicting with peripheral Y; you can use the alternative pin Z to resolve this."

3. Code Explanation and Generation for Live Code Preview: With an AI button integrated into the new Live Code Preview window, the generated HAL2/LL code could be explained line by line. Furthermore, by prompting "Write a DMA receive function template for this UART configuration", the assistant could generate driver snippets tailored exactly to the current setup.

4. Chat-Based Dynamic Documentation: The new online documentation portal is a great step, but instead of drowning in thousands of pages of Reference Manuals or Datasheet PDFs, it would be much more efficient to ask the integrated LLM, "How does this specific timer register work?" and get an instant summary and usage example.

Since CubeMX2 and next-gen MCUs (like STM32C5) are turning a new page, it is the perfect time to equip the tool with a true "AI Assistant". I hope we can see these suggestions on the roadmap in the upcoming quarterly releases. I'll be following the updates with excitement!

[Etienne_Parmentier]

Please, please less AI BS. No one likes it. It does not work in a reliable way. I would instead focus on better documentation and a better API to make things easier to learn by humans.

[mrkozmic]

I'm currently testing AI for development on STM32. I have two boards, one with Zephyr and with one-bare metal. I use CubeIDE for setup of the bare-metal application, but I let AI write the code in vscode. 
The goal is to write/read as little code as possible (for this test only). 
I have so far got Zephyr setup with LVGL and a few nice screens. I have not written a single line of code!
In the same application, AI also wrote me a bare-metal driver for an encoder and a stepper motor for a CNC. 
I would love to have support for Codex in CubeIDE just like in vscode.
It have saved me a few days work already.