Meta releases Muse Wonder, its first frontier AI model after 2025 overhaul

Meta said on 12 October 2026 that it had released Muse Wonder and Muse Wonder Full, its first frontier-class artificial intelligence models since Scale AI founder Alexandr Wang restructured the company's AI division in August 2025, TechCrunch reported. The standard Wonder model costs $4 per million input tokens and $20 per million output tokens through Meta's commercial API, roughly matching the price of OpenAI's GPT-6 Sol and Anthropic's Claude Opus 5.5, according to TechCrunch.

What the launch covers

• Muse Wonder and Muse Wonder Full are Meta Superintelligence Labs' first frontier model release since the August 2025 reorganisation, which Meta's own research blog said began "fourteen months ago."
• Meta said in its announcement that the two models "mark the next step on our scaling ladder and a major leap toward personal superintelligence."
• Both models were trained at Meta's Prometheus data centre in central Ohio, which came online earlier in 2026, according to TechCrunch's review of Meta's announcement.
• Meta built in a feature it calls Active Memory, described in its research blog as "carrying important context about you across conversations and actively retrieving the details relevant to the task at hand."
• Citing Meta's own testing, TechCrunch reported that Muse Wonder "now scores higher on MemoryBench than Fable, Opus, or Astra" — Meta's shorthand for rival models Claude Fable 5.1, Claude Opus and GPT-6 Astra.
• Wonder is available now through Meta's commercial API and its Muse Code programming assistant; Wonder Full is reserved for paying Muse Code subscribers and "will not be served through the API," TechCrunch reported.
• Research scientist Kavya Kopparapu said in a video accompanying the launch that the team "put a tremendous amount of effort into making sure your model can remember what's important to you over days, weeks, months, and beyond," according to TechCrunch.

Why it matters

The release ends more than a year in which Meta's restructured AI labs shipped only the lighter Muse Spark model and a Muse consumer agent while OpenAI, Anthropic and Google released successive frontier systems, according to TechCrunch. By pricing Wonder level with Claude Opus 5.5 and GPT-6 Sol rather than undercutting them, Meta is for now competing on capability rather than cost — a pricing pattern that companies and public institutions in the Caspian region evaluating AI vendors will need to track as the field of comparably priced frontier models widens.