Meta has introduced a new AI model, Muse Glimmer, now available for download and operation on personal computers. The model, a reduced version of Muse Spark 1.2, is designed to perform tasks efficiently even on devices with minimal hardware requirements.
The Muse Glimmer model is equipped with 30 billion parameters and can operate on a Mac or PC using only one graphics card. This allows it to run without the need for a constant internet connection or cloud services. According to Meta, Muse Glimmer is optimized for performing agent tasks such as file organization and schedule management, making it an attractive tool for a wide range of users.
In addition to download availability, Muse Glimmer allows users to adapt the model to their specific needs. This became possible by increasing the learning efficiency of less voluminous models, demonstrating excellent performance without loss of quality.
The launch of Muse Glimmer takes place against the backdrop of active discussions in the United States about the availability of powerful AI models. Meta founder Mark Zuckerberg advocates for their wider dissemination, believing they should not be concentrated exclusively within large companies. Representatives from OpenAI and Anthropic express concerns about the safety of open models, as AI systems have previously exceeded their intended environments. Along with Meta, companies like Google, DeepSeek, Moonshot, and Alibaba are also working on similar models, intensifying competition in the artificial intelligence field.
The importance of developing local AI models is growing, especially considering previous issues with Muse Spark 1.1 when a test environment provided access to systems of another company. The optimization of Muse Glimmer for local use may lay the foundation for a safer and more controlled artificial intelligence environment.
