Meta releases Muse Glimmer and open-weights Muse Spark 1.2
Meta’s new 30B Muse Glimmer model runs locally—full Muse Spark 1.2 weights open-sourced.
Meta is doubling down on open, locally deployable AI with the release of Muse Glimmer, a 30-billion parameter open-weight model optimized for running AI agents directly on consumer GPUs like those in high-end laptops or desktops. The model is designed to be lightweight enough for edge deployment while maintaining strong reasoning and agentic capabilities, signaling Meta’s commitment to bringing advanced AI to everyday hardware.
In a parallel move, Meta CEO Mark Zuckerberg announced the open-sourcing of Muse Spark 1.2 weights, a more powerful sibling to Muse Glimmer that’s intended for research and enterprise use. By open-weighting both models, Meta is accelerating community adoption, enabling developers to fine-tune, audit, and deploy agentic AI without reliance on proprietary cloud APIs. This aligns with Meta’s broader vision of 'personal superintelligence'—putting capable AI directly into users’ hands, not just data centers.
- Muse Glimmer is a 30B open-weight model optimized for local AI agent deployment on consumer hardware
- Meta open-sourced Muse Spark 1.2 weights, enabling research and enterprise use without cloud dependency
- Part of Meta’s push for 'personal superintelligence'—bringing agentic AI to everyday devices
Why It Matters
By open-weighting models like Muse Spark 1.2 and Muse Glimmer, Meta is democratizing agentic AI, shifting power from cloud providers to users and developers.