Meta launches Muse Spark Open, the most capable US open-weight model with 816 B total, 42 B active parameters and a 1 M‑token context (Meta AI Research)
agents meta
| Source: Techmeme | Original article
Meta has launched Muse Spark Open, a US open‑weight model with 816 billion total and 42 billion active parameters and a 1 million‑token context window, bringing advanced agentic AI to desktops.
Meta AI Research announced today the launch of Muse Spark Open, billed as the most capable open‑weight model released in the United States to date. The new system packs 816 billion total parameters, of which 42 billion are active, and supports a 1 million‑token context window. Designed for “coding and agentic work,” Meta says the model excels at long‑horizon tasks, complex software engineering and multi‑step reasoning.
Muse Spark Open is distributed under a limited‑commercial‑use license and runs on Apple’s M5 Ultra Mac Studios. The partnership with Apple enables the model to operate on desktop hardware, though Meta notes that a full deployment demands roughly 1 TB of RAM. By making the weights publicly available, Meta aims to lower the barrier for developers to experiment with large‑scale, agentic AI on personal machines, a capability previously confined to cloud‑based services.
The release matters for several reasons. First, it marks a shift toward openly accessible, high‑parameter models in the U.S., counterbalancing the dominance of closed‑source offerings from rivals. Second, the model’s multimodal, personal‑assistant ambitions—ranging from environment analysis to wellness support—signal Meta’s push toward “personal superintelligence” that can run locally. Finally, the hardware demands highlight the growing divide between cutting‑edge AI research and the practical resources needed for everyday users.
What to watch next includes the uptake of Muse Spark Open among developers, performance benchmarks against competing models, and any extensions of the licensing terms. Observers will also monitor how Apple’s hardware roadmap evolves to accommodate such massive models, and whether regulatory scrutiny intensifies as open‑weight AI becomes more powerful and widely deployable.
Sources
Back to AIPULSEN