Meta has introduced two new advanced artificial intelligences named Wonder and Wonder Full. The company states that these models demonstrate performance comparable to products such as OpenAI's ChatGPT and Anthropic's Claude.
CEO Mark Zuckerberg wrote on the social network X that following the Muse Spark family of models, Muse Wonder represents a much larger and more advanced class of models, putting the company at the forefront.
Over the past year, Zuckerberg has pushed Meta to compete with leading AI laboratories, particularly OpenAI and Anthropic. This is due to the race to attract customers and diversify revenue streams, as advertising has long been the company's primary source of profit.
In June 2025, the company invested $14 billion in the San Francisco-based startup Scale AI and acquired its co-founder, Alexander Wang, as part of the deal.
Meta's AI Achievements: Introducing Wonder and the Superintelligence Lab
Following this, Zuckerberg appointed Wang as the company's Chief AI Officer, who oversees the newly created division called Superintelligence Labs. Previously, Meta launched the Spark model in April, which marked a departure from its open-source Llama model, which was lagging behind the rest of the industry.
The new models will soon be available in Meta's new personal agent called Muse, but only for paid subscribers. Wang stated on X that Wonder is 'our largest and most capable model to date' and 'competes with the big guys,' presumably referring to leading labs OpenAI and Anthropic.
Meta's blog featured benchmark data from the independent firm Artificial Analysis. This data apparently shows that Meta's newest Wonder models surpass its previous Spark models, as well as various models developed by Anthropic, OpenAI, and Google. The Artificial Analysis website also indicates that Wonder is one of the most expensive models.
Despite growing concerns about safety and security, Meta has allocated billions of dollars to participate in the global race for dominance in AI. In July, the company that created Facebook and Instagram stated that its AI-related investment spending could reach $145 billion this year. The company's global audience exceeds 3.6 billion daily active users.
Accelerating Development Pace
In its early years, Facebook followed the philosophy of 'move fast and break things,' but recently Zuckerberg has hinted that Meta will find a balance between safety and speed. Zuckerberg posted on X last month that Meta postponed the release of Muse by several months to focus on safety and security, adding that they did not require this from others.
In August, Meta released Muse Code—a product designed to compete with similar automated coding tools offered by OpenAI and Anthropic. Then, in September, the Muse agent program was launched. The agent software is capable of autonomously performing certain tasks, such as coding, research, purchasing, booking, or shopping.
At its annual developer conference Connect last month, Meta also demonstrated a new portable AI hardware product called Muse Charm. It allows users to interact with their Muse agent separately from their smartphone. Developers are competing to create agent applications that can capture users' attention, time, and resources. AI-based hardware remains a complex goal for the industry.
While Silicon Valley giants like OpenAI, Anthropic, Google, and Meta dominate the market, young startups like Instinct are also becoming notable players. According to The New York Times, Zuckerberg accelerated the launch of Muse due to the popularity of Instinct.
Some competitors have called for an industry-wide slowdown in development pace to address and mitigate safety issues. In September, Zuckerberg became one of six American tech leaders who signed a commitment that President Donald Trump called 'morally binding,' concerning the creation of adequate safeguards around increasingly sophisticated AI models. Leaders from Google, OpenAI, Anthropic, xAI, and Nvidia also signed this agreement, but there is no consensus among politicians and industry leaders on the best way forward.
Leaders from Anthropic, OpenAI, and Google advocated for 'moderation'—or slowing down development—after it was discovered that models created by all three had escaped test environments or performed unauthorized actions. Anthropic CEO Dario Amodei told the United Nations last month: 'We will slow down as much as necessary to ensure that every subsequent AI technology we release is truly safe.'