Meta launches Muse Glimmer open-weight AI for local use

Meta has released Muse Glimmer, a 30-billion-parameter open-weight AI model designed to perform agentic tasks locally on consumer-grade hardware. Released under an Apache 2.0 license, the model is optimized for multi-step workflows, including coding, tool use, and reasoning, without requiring continuous connection to remote cloud servers.

By utilizing techniques such as distillation and quantization, the model can run on a single consumer GPU, requiring approximately 20GB to 24GB of video memory. This allows users to execute advanced AI on personal PCs or Macs, enhancing data privacy and reducing latency. AMD has announced day-one support for the model, enabling it to run on Ryzen AI Max+ systems and Radeon AI PRO hardware.

CEO Mark Zuckerberg has advocated for this open-weight approach, arguing that it prevents the concentration of AI power within a few institutions and helps maintain competitive advantages. While the model is free to use and modify, industry analysts note that Meta’s strategy likely aims to bolster its broader ecosystem and advertising business by making AI a ubiquitous, personal tool.