SKYNET://COUNTDOWN SYS:MONITORING ▮

Meta Superintelligence Lab Launches Muse Wonder Frontier Models with Persistent 'Active Memory'

[Commercial Release]

Meta's Superintelligence Lab released Muse Wonder and Muse Wonder Full, its first frontier models, which it says match competitors like Fable 5.1 and GPT-6 Astra on benchmarks and lead on the MemoryBench continual-learning test. Wonder is priced at $4/$20 per million input/output tokens via API, while Wonder Full is restricted to Muse Code subscribers and both power the Muse agent. The models were trained at Meta's new Prometheus data center in Ohio, following Alexandr Wang's restructuring of Meta's AI efforts.

Risk: [+0.01% ↑] [0 days]
AGI: [+0.01% ↑] [0 days]
> Impact_Analysis

Skynet Chance (+0.01%): Adding another well-resourced frontier competitor explicitly pursuing 'superintelligence' intensifies race dynamics, and long-horizon agents with persistent memory expand autonomy surfaces that are harder to oversee. The article does not mention accompanying safety measures.

Skynet Date (+0 days): Meta's entry at the frontier, backed by massive new compute infrastructure, increases competitive pressure that may speed capability deployment ahead of safety work.

AGI Progress (+0.01%): Claimed state-of-the-art results on memory and continual-learning benchmarks address a key known limitation on the path to general agents. Overall benchmark performance, however, roughly matches existing frontier models rather than surpassing them.

AGI Date (+0 days): A fourth major lab reaching frontier parity on its new Prometheus data center, while explicitly 'climbing a scaling ladder,' adds competitive momentum and compute that could modestly accelerate AGI timelines.

>> Read the original story at TechCrunch

<< All AI news for October 12, 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