@finkd
Post
Today we're also opening the weights for Muse Glimmer, a great 30B parameter dense model that can run locally. Soon we'll also release the weights for Muse Spark 1.2, our latest foundation model. Meta is a strong supporter of open source and I'm proud of these releases. Congrats to @alexandr\_wang and the MSL team for all your great work on these models.
Explanation
What it says: Mark Zuckerberg says Meta is releasing the weights for Muse Glimmer, a 30B-parameter dense model designed to run locally, with Muse Spark 1.2—described as Meta’s latest foundation model—coming soon. He frames both releases as evidence of Meta’s support for open source and credits Alexandr Wang and the MSL team.
Context: The post gives no benchmarks, architecture details beyond “30B dense,” hardware requirements, license terms, context length, training data, or intended task specialization. “Opening the weights” also does not by itself establish that the model is fully open source under a permissive license.
Why it matters: A competent 30B dense model with downloadable weights could be attractive for local inference, private deployments, experimentation, and fine-tuning, especially on multi-GPU systems. The important follow-up is the actual release: verify license restrictions, benchmark quality, quantized-weight availability, VRAM/RAM needs, inference-stack support, context length, and whether Glimmer materially beats existing open-weight ~30B models before investing time in it.