Summary
Meta released Muse Spark Open on October 15, a mixture-of-experts model with roughly 816B total parameters and about 42B activated per token, featuring a 1 million token context and native multimodality, billed as the strongest US model among open-weight releases. Weights can be downloaded from Hug…
Key points
- For the first time, developers can access near-frontier multimodal coding and agent capabilities via open weights locally or in their own cloud, controlling costs and data themselves.
- An open-source frontier model releases top-tier coding agent capabilities from a handful of cloud providers, directly reshaping enterprise AI procurement and self-hosted inference options.
- Enterprises can evaluate replacing some cloud API spending with self-hosted inference, while dev teams must re-compare the cost and compliance terms of open versus closed commercial models.
Editorial note
This page is Code & Chain's editorial summary of public sources. It may be prepared with AI assistance and published through an automated workflow. Refer to the original sources; this content is not investment, legal, or tax advice.