Meta opens weights for Muse Glimmer and announces Muse Spark 1.2 release

Meta releases Muse Glimmer, a 30B open agentic AI model that runs locally on PCs

Meta releases Muse Glimmer, a 30B open-weight agentic model running on consumer GPUs (4-bit <20GB), beating Gemma4 and Qwen3.6. Local agents eliminate cloud dependency—cutting API costs and latency for teams prioritizing data control over vendor lock-in.