Muse Glimmer: Meta open-sources a 30B agentic LLM tuned for local AI workflows

Meta Superintelligence Labs has released Muse Glimmer, a 30-billion-parameter open agentic model built for always-on local agent workflows. The weights land on Hugging Face today under a permissive Apache 2.0 license, alongside developer documentation for building and running your own agents. It is small enough to run on a Mac or PC with a single consumer GPU, so agents can live on your machine rather than in the cloud.

The model targets use cases that demand AI anywhere, anytime, with or without a network connection: local agents, function calling, local coding, and LLM-as-a-judge evaluation. Meta says Muse Glimmer delivers strong performance on key agentic use cases and benchmarks compared with leading models in its size category. Running locally also keeps data on-device, which matters for privacy-sensitive and always-on workloads.

The release continues Meta's tradition of sharing fundamental AI research, and lands as the industry shifts toward smaller, more efficient models that can operate independently of cloud infrastructure. The Hacker News community responded fast: the announcement climbed to the top of the front page within hours with heavy discussion. Developers can pull the weights from Hugging Face today and have a local agent running on their own hardware by the end of the day.