Meta came out swinging with its newest AI, Muse Wonder, calling it the company's most capable model yet and a step toward "personal superintelligence." The claims are big. The reality, as usual, is more nuanced.
What Meta actually launched
Meta released Muse Wonder and a stronger version called Muse Wonder Full, which Mark Zuckerberg framed as a step toward personal superintelligence. The models add features like Active Memory, which carries useful information across conversations, and Active Focus, which helps the AI stay on long, difficult tasks. On the Artificial Analysis Intelligence Index, Wonder Full scored 56 and standard Wonder scored 52, both at maximum reasoning effort.
Does it really beat ChatGPT and Claude?
Here is where the headline needs a closer look. Wonder Full's score of 56 landed above OpenAI's GPT-6 and Anthropic's Claude Fable 5.1, which both scored 53. But it came in below Claude's strongest model, Opus 5.5, which scored 58. So Meta can fairly say it beat some top models, but not all of them, and not Claude's best.
Not everyone is convinced even of that. Bindu Reddy, CEO of Abacus.AI, said publicly that Muse Wonder was nowhere near the level Meta claimed, and promised more detailed results would follow. Benchmarks are also only one view of a model, and real-world performance can look different from a leaderboard.
Price is part of the story too. Standard Wonder is competitive at $4 per million input tokens, matching Claude Opus 5.5 and undercutting pricier rivals. But Wonder Full, the version that posts those benchmark numbers, is only available through Meta's paid Muse Code subscriptions, which run $50 or $200 a month. These are Meta's most expensive models to date.
What it means for you
More competition at the top is good news: it usually means better tools and pressure on prices. But it also means you should read "beats everyone" claims with a healthy dose of skepticism. The best model for your team depends on your actual tasks and budget, not on who is leading a single benchmark this week.
Muse Wonder is a genuine step up for Meta and more proof that the top of the AI market is getting crowded and competitive. Just take the "beats everyone" framing with a grain of salt, and judge any model on your own work, not one score.
