STMicroelectronics, a global semiconductor leader, has expanded its STM32 AI Model Zoo with new models and improved project support to accelerate the prototyping and development of embedded AI applications. This enhancement strengthens what is already the industry’s largest collection of vision, audio, and sensing models designed for integration into devices such as wearables, smart cameras, advanced sensors, security and safety systems, and robotics.
Transforming data science into a fully optimized embedded application is a complex task, and developers need end-to-end support,” said Stephane Henry, Edge AI Solution Group VP at STMicroelectronics. “With STM32 AI Model Zoo 4.0, we are not only expanding the range of available models to help the STM32 developer community accelerate their projects, but also reinforcing the entire infrastructure from development to deployment. This reflects our commitment to making Physical AI a practical reality.
Embedding AI into everyday accessories, appliances, and other electronic devices delivers transformative benefits, boosting efficiency and reducing energy consumption. However, these products often rely on compact microcontrollers with limited processing power and memory, making it challenging for developers to balance AI performance with efficiency. ST’s latest AI Model Zoo helps overcome these constraints by providing optimized models that enable designers to make the most of available resources and build highly efficient, low-power AI applications.
This Model Zoo is part of the ST Edge AI Suite, a comprehensive set of tools, libraries, and utilities designed to simplify and speed up the development and deployment of AI algorithms on ST hardware. It ensures a smooth, efficient transition from prototyping to full-scale production.
For more than a decade, ST has led research, innovation, and development in edge AI, empowering developers to navigate the challenges of deploying AI at the edge using both software- and hardware-accelerated models. Today, ST’s AI tools support more than 160,000 projects each year, underscoring the company’s commitment to advancing edge intelligence at scale.
The STM32 family from ST is the world’s most widely adopted microcontroller platform, powering applications ranging from consumer appliances and wearables to communication infrastructure, smart grids, smart cities, industrial automation, and even low-earth-orbit satellites. By enabling AI deployment on general-purpose MCUs across these diverse sectors, ST brings advanced technology to end users quickly and cost-effectively, while also contributing to greater sustainability.
Building on its leadership in embedded AI, ST is expanding its portfolio with AI-accelerated microcontrollers, including the STM32N6 series. These advancements position the company at the forefront of the rapidly growing embedded and edge AI market.
Technical Highlights
- Doubles the model families from 30 to 60, expanding the library to over 140 ready-made MCU models, the largest collection in the embedded AI industry.
- Evolves from a model catalog to a complete workflow, offering scripts that support training, optimization, and seamless integration with application libraries for maximum performance and efficiency.
- Adds native PyTorch model support, enhancing compatibility alongside existing frameworks including TensorFlow Lite, Keras, LiteRT, and ONNX.
- Boosts embedded AI efficiency with advanced compression techniques and sub-byte quantized models for ultra-low-power deployment.
