Meta launches Muse Wonder, claiming it beats GPT-6 and Gemini 4 on benchmarks
Meta's new Muse Wonder model outscores GPT-6, Gemini 4 and Claude Fable 5.1 on independent leaderboards, ranking behind only Opus 5.5.
Mark Zuckerberg's AI ranked 32nd last year and now it's beating OpenAI and Google In April 2025 Meta showed off Llama 4 sitting in second place on a public AI leaderboard But nobody outside Meta could download the version on the leaderboard The real Llama 4 came in 32nd So Zuckerberg started spending He put $14.3 billion into a deal to bring in a new head of AI and offered some researchers pay packages reportedly worth hundreds of millions Then Meta spent 14 months rebuilding its AI from scratch Today it launched Muse Wonder Its full version scores higher than GPT-6 and Gemini 4 and even Claude Fable 5.1 on the main independent AI rankings Only one model still scores higher and that's Claude Opus 5.5 Say what you want about Zuckerberg but this is the fastest comeback I've seen in tech https://x.com/AIatMeta/status/2109691148943999167/video/1
This post quotes another post: x.com/finkd/status/2109689305073614918