Meta releases Muse Glimmer as a compact open-weight model for local agentic work

a report, one report and reports report that Meta released Muse Glimmer, an open-weight model designed for smaller reasoning and agentic workloads on a single GPU or personal computer. Meta also signalled a later open-weight Muse Spark 1.2 release. Detailed model-card, benchmark and licence evidence was not located in the accessible official newsroom material.

Why it matters: Efficient local models can reduce inference cost and data exposure. Impact: Enterprises should validate licence, memory footprint, tool-use security, accuracy and total cost on their own workloads rather than rely on vendor comparisons.

a report · press reports · a media report