Meta Unveils Muse Glimmer: AI Locally Powered Without the Cloud

Meta has introduced Muse Glimmer, a 30-billion-parameter AI model designed to operate locally on consumer devices, reducing dependency on cloud infrastructure. This model supports a range of applications, has enhanced memory efficiency, and is available under an open-source Apache 2.0 license for developer customization.

In a significant stride towards decentralizing artificial intelligence, Meta has unveiled Muse Glimmer, an innovative AI model developed to function locally on consumer computers. Designed by Meta Superintelligence Labs, this 30-billion-parameter model empowers developers to create AI-driven tools without constantly relying on cloud services or an internet connection.

Muse Glimmer is tailored for AI agents capable of performing tasks independently, such as managing schedules, drafting messages, and organizing files. This model also supports developer applications, including coding, by processing both text and image data from over 100 languages, enhancing its versatility and utility.

One notable challenge of running sizable AI models locally is their substantial memory requirement. To make Muse Glimmer accessible to more users, Meta employed quantization to compress it to under 20 GB, enabling it to operate within 24 to 32 GB of memory. This development marks a leap forward in AI efficiency, offering real-time capabilities fully operational on user devices.