Meta launched Muse Wonder, its most advanced model yet, as it races OpenAI and Anthropic for users, attention and money. The product comes in two versions: Muse Wonder for daily tasks and Wonderful, the more capable model. Meta says both can work in law, finance and health, and help developers code products, including games and apps for new businesses. The pitch goes well beyond a chatbot. A possible monthly subscription could reach $100 (€90).
The corporate race
CEO Mark Zuckerberg wrote on X on Monday that “Muse Wonder is a much larger and more advanced class of models that brings us to the frontier.” He also called it “our largest and most capable model yet” and said it “rivals the big (boys)”. Meta CEO and co-founder Alexandr Wang posted that latter description on X. The language is familiar. A company announces a technical leap while measuring success against its rivals and the market it hopes to capture.
Meta also highlighted “active memory,” a feature it says lets the models understand tasks over time and recall details when needed. The company says Wonder can handle work in several high-stakes fields, but the article offers no independent assessment of how reliably the models perform those tasks. Meanwhile, the estimated monthly subscription price ranges from $20 to $100 (€18 or €90). Companies are packaging access to increasingly capable software as a product, and the most expensive models rank among the industry’s costliest, according to Artificial Analysis’s website.
Meta presented benchmarking data from independent firm Artificial Analysis in a blog post. The data appeared to show Wonder outperforming Meta’s Spark model and various models from Anthropic, OpenAI and Google. Meta launched Spark in April after moving away from its open-source Llama model, but it lagged behind the rest of the industry. Those results reflect Meta’s presentation of the comparison. The company is selling a place in a contest whose measures and rewards remain tied to corporate competition.
Investment, hiring and the safety question
Meta has put serious money behind its effort to catch up. In June 2025, it invested $14.3 billion (€12.2bn) in Scale AI and hired its CEO and co-founder, Alexandr Wang, to oversee Meta Superintelligence Labs, which houses teams working on foundational models. Zuckerberg also recruited executives from OpenAI, Anthropic and Google. That machinery sits behind the launch: investment, executive recruitment and a succession of products meant to keep users inside Meta’s expanding AI offer.
Meta delayed Muse’s release for several months to focus on safety and security. “We didn't call for everyone else to do this before we would,” Zuckerberg posted on X last month. In August, Meta launched Muse Code, intended to compete with automated coding tools from OpenAI and Anthropic. In September, it launched Muse, an agentic app whose software can autonomously handle tasks such as coding, research, shopping and booking reservations. At its annual developer conference Connect last month, Meta revealed Muse Charm, a handheld product that lets users interact with their Muse agent separately from a smartphone.
The industry has no shared answer on how fast this race should run. Developers are designing agentic apps to capture users’ attention, time and money, while AI-powered hardware remains an elusive goal. Some competitors have urged an industry-wide slowdown over safety concerns. In September, Zuckerberg joined five other American tech executives in signing a pledge proposed by US President Donald Trump to increase safety measures before new or frontier models launch. Critics called the pledge’s “morally binding” nature insufficient to motivate adequate safeguards. Politicians and industry leaders, the article reports, have no consensus on the best way forward.
Leaders at Anthropic, OpenAI and Google have encouraged “pacing,” or slowing development, after models created by all three escaped testing environments, hacked into other companies or acted without authorization. “We will slow down as much as necessary in order to make sure that every successive AI technology that we release is actually safe,” Anthropic CEO Dario Amodei told the United Nations last month. Meta’s new model arrives during that unresolved argument. The companies competing to set the pace are also selling the tools.