Meta’s claimed Muse Wonder launch and its “caught up with OpenAI and Anthropic” claim: Meta announced Muse Wonder and Muse Wonder Full on October 12, 2026, but the cited material does not establish Zuckerberg’s exact “caught up” statement or validate the viral rival-model comparisons.

What Meta announced

Meta describes both models as frontier systems for personal and professional agent tasks. Muse Wonder is available through Meta Model API and Muse Code; Muse Wonder Full is available to Muse Code subscribers. Meta said Muse Wonder would soon power its Muse personal-agent product, initially for subscribers. Meta AI’s announcement

What the benchmark evidence says

Artificial Analysis reported Intelligence Index v4.3 scores of 52 for Muse Wonder and 56 for Muse Wonder Full at maximum reasoning effort, compared with 48 for Muse Spark 1.3. It reported gains for Muse Wonder Full across all ten evaluations, including Terminal-Bench 4.0, GDPval-AA v2.1 and AA-Briefcase. These results support a measurable improvement over Meta’s prior model, but do not by themselves prove an overall lead over OpenAI or Anthropic. Artificial Analysis’s results

Safety and the other claims

Meta says the models were trained for robustness to adversarial instructions and careful handling of private information, and describes safeguards against chemical, biological, radiological, nuclear and cyber misuse. Its launch post says a fuller safety evaluation report is forthcoming. Meta’s research announcement The supplied posts’ claims that the model beats “Claude Opus 5.5” and “GPT-6 Astra,” costs $200 per month, accompanies increased infrastructure spending, or reflects Zuckerberg rejecting a slowdown are not substantiated by the primary launch materials or benchmark evidence cited here. The sources do not name an independently verified comparison showing Muse Wonder has caught up with OpenAI and Anthropic.

Meta announcement · Artificial Analysis results · Meta AI availability post

Research trace 4

The supplied posts attribute a frontier-model launch and specific capabilities, prices, products, spending, and safety positions to Meta and Mark Zuckerberg. These independent reads check the posts, Meta's public account, current news reporting, and whether the named model appears in public reporting.

The first round found Meta team members announcing Muse Wonder and Meta's stated availability, alongside unverified repost claims about rival benchmarks, pricing, spending, and safety. These independent reads check Meta's launch materials and look for benchmark evidence and reporting on the remaining claims.

The previous results establish that Meta announced Muse Wonder and Muse Wonder Full, their availability, and Meta's own safety and capability claims. They also show Artificial Analysis benchmark scores, but do not establish the viral comparisons to GPT-6 Astra or Claude Opus 5.5, the $200 price, infrastructure-spending claims, or Zuckerberg's precise remarks; this step checks for direct reporting and primary-source detail on those points.

The earlier steps found Meta’s official launch and availability statements, Meta’s safety claims, and Artificial Analysis benchmark results, but no reliable support for the viral claims about a $200 tier, spending, or Zuckerberg’s precise remarks. The reviewed independent and primary sources do not resolve those remaining claims.