Multiverse Releases Compressed AI Models for IoT Devices

What's Happening?

Multiverse Computing, a European AI startup, has introduced two highly compressed AI models designed for use in Internet of Things (IoT) devices. These models, named SuperFly and ChickBrain, are part of Multiverse's 'Model Zoo' and are intended to operate on devices with limited processing power, such as smartphones and home appliances. SuperFly, a compressed version of Hugging Face's SmolLM2 135 model, has 94 million parameters and is designed for simple voice command tasks. ChickBrain, with 3.2 billion parameters, is a more advanced model capable of reasoning and is a compressed version of Meta's Llama 3.1 8B model. Multiverse's compression technology, CompactifAI, allows these models to maintain high performance despite their reduced size.

Why It's Important?

The introduction of these compressed models by Multiverse Computing represents a significant advancement in AI technology, particularly for IoT applications. By enabling AI models to run efficiently on devices with limited computational resources, Multiverse is expanding the potential for AI integration in everyday consumer electronics. This could lead to more intelligent and responsive home appliances, enhancing user experience and convenience. Additionally, the ability to run AI models locally on devices without internet connectivity addresses privacy concerns and reduces reliance on cloud computing, which can be costly and less secure.

What's Next?

Multiverse is in discussions with major device manufacturers, including Apple, Samsung, and Sony, to integrate these models into their products. The company also offers its models via an API on AWS, allowing developers to access and utilize these compressed models for various applications. As the demand for AI-driven IoT solutions grows, Multiverse's technology could become a standard in the industry, prompting other companies to develop similar compression techniques to remain competitive.