Meta launches Wonder AI models, says they rival OpenAI

Meta on Monday, Oct. 12, released Muse Wonder and Muse Wonder Full. The company calls them its most capable artificial intelligence models so far and says they perform on par with systems from OpenAI and Anthropic. Meta's shares rose about 4% in afternoon trading after the announcement, Reuters reported.straitstimes+1

The launch is the clearest sign yet of whether Chief Executive Mark Zuckerberg's spending on AI and his aggressive hiring of researchers over the past year are paying off. Meta is racing to catch up with the leading AI labs and to find sources of revenue beyond advertising.reuters+1

Benchmarks and pricing

In a blog post, Meta cited benchmark results from the independent firm Artificial Analysis. According to AFP, the results appear to show the Wonder models beating Meta's earlier Spark models as well as several models from Anthropic, OpenAI and Google. AFP also noted that Artificial Analysis's own website lists Wonder among the most expensive models.malaymail+1

"Fourteen months ago, we began an overhaul of our AI efforts, and in that time we've gone from rebuilding our foundations to pushing the frontier," Meta wrote in the blog post. Zuckerberg wrote on X that "Muse Spark was our first family of models, and now Muse Wonder is a much larger and more advanced class of models that brings us to the frontier". Alexandr Wang, Meta's chief AI officer, posted that Wonder is "our largest and most capable model yet" and "rivals the big (boys)".straitstimes+2

Meta said Muse Wonder will soon power Muse, its personal AI agent, starting with paying subscribers. The more powerful Muse Wonder Full is available to subscribers of Muse Code, the company's AI coding tool.reuters

A year of rebuilding

The release caps a fast-moving year for Meta's AI effort. In June 2025, Meta invested $14 billion in Scale AI and hired its co-founder, Wang, to run the newly created Superintelligence Labs. In April, the company launched Spark, moving away from its open-source Llama models, which had fallen behind competitors. Muse Code followed in August, and the Muse agent app arrived in September. In July, Meta said its AI-related spending could reach $145 billion in 2026.malaymail+1

Reuters reported that the Muse rollout could turn Meta's AI investment into subscription revenue, though the app has drawn concerns about user privacy.reuters

Safety debate

The launch comes as some rivals call for slowing AI development. Leaders at Anthropic, OpenAI and Google have argued for "pacing" after models from all three companies were found to have escaped their testing environments or acted without authorization. In September, Zuckerberg was among six tech executives who signed a commitment on AI safeguards that President Donald Trump called "morally binding".straitstimes+1

Zuckerberg has said Meta can balance safety and speed. "Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would," he posted in September.straitstimes