SKYNET://COUNTDOWN SYS:MONITORING ▮
Skynet Countdown Logo SKYNETCOUNTDOWN

Meta Releases Muse Spark Open, an 816B-Parameter Open-Weight Model Capable of Running Locally on Macs

[Commercial Release]

Meta has launched Muse Spark Open, an 816-billion-parameter, natively multimodal open-weight model with a 1 million token context window, which it calls the most capable open-weight model in the US. The model is built for complex coding and agentic tasks, is designed to run locally on high-memory Macs such as the M5 Ultra Mac Studio, and has been demonstrated on agentic workflows like app development and video editing. It is available now through several API platforms.

[Meta] [Multimodal] [Agentic AI] [local inference] [open-weight models]
Risk: [+0.01% ↑] [0 days]
AGI: [+0.01% ↑] [0 days]
> Impact_Analysis

Skynet Chance (+0.01%): Releasing a highly capable agentic, multimodal model with open weights makes it harder to enforce safety controls or retract the model if misuse or dangerous behaviors emerge. Hardware requirements of at least 1TB of RAM somewhat limit how widely it can be run locally.

Skynet Date (+0 days): Wider access to frontier-level agentic capabilities modestly accelerates diffusion of powerful autonomous systems, though this is an incremental step rather than a capability leap.

AGI Progress (+0.01%): The model shows solid but incremental progress in agentic coding and multimodal task execution, such as full app builds and video editing, which narrows the gap between open and closed models.

AGI Date (+0 days): Open weights let researchers and developers build on frontier-scale agentic capabilities, slightly speeding the overall pace of capability development across the ecosystem.

>> Read the original story at TechCrunch

<< All AI news for October 15, 2026

[ Get the daily index digest on Telegram → ] One post a day at 17:00 UTC: how the Skynet Chance and AGI Progress numbers moved, and the news that moved them. Nothing else.

Related AI News