Meta has introduced Muse Spark Open, an open-weight model with 816 billion parameters and a 1 million token context window.
Designed for local execution on macOS, the model is optimized for complex programming and agent-based tasks.
We examine the hardware requirements, which necessitate at least 1 terabyte of RAM, and explore how Meta tested the model's performance in environments like Muse Code and DaVinci Resolve.
Capabilities and performance
Muse Spark Open has 816 billion parameters and can handle a huge amount of information, with a memory “window” of 1 million data units. It is capable of undertaking complex programming tasks and performing autonomous actions. Meta tested its capabilities in the Muse Code programming tool, as well as in the DaVinci Resolve video editing software.
Requirements for local use
Although the program is designed for Mac computers, its local operation requires extremely powerful hardware. To be able to run on your computer, at least 1 terabyte of RAM is required. According to the company, the best performance is achieved on a Mac Studio with the M5 Ultra processor.
For those who do not have such equipment, the model is already available through various other platforms, offering a more affordable and transparent alternative for developers and companies.
G
Hacks.gr on Google
Set it as a preferred source for cybersecurity updates.
Preferred source
Top comments (0)