According to AI at Meta, the Muse Glimmer 30B open-weight model runs at practical speeds on a single consumer GPU at 4-bit precision and leads peers in tool use and multi-step planning.

According to TheRundownAI, Meta open-sourced Muse Glimmer 30B and plans Muse Spark 1.2 open weights, signaling a push for distributed superintelligence.

According to AIatMeta, Muse Glimmer is a 30B open-weight model for local agent workflows, Apache 2.0 licensed, optimized for Macs and PC GPUs.