Meta unveils Muse Wonder models in fresh push into frontier AI

Meta has launched two new artificial intelligence models, Muse Wonder and Muse Wonder Full, as the company steps up efforts to compete with OpenAI and Anthropic. Meta's Superintelligence Lab unveiled the models on Monday and says they are the most powerful AI systems it has yet created. Muse Wonder also represents Meta's first major step toward what is often referred to as a frontier model, directly facing some of the most advanced models created by other AI firms. The launch follows a year of substantial investments by Meta's CEO Mark Zuckerberg in bolstering the company's AI efforts, such as recruiting AI researchers from rival companies and investing in computing technology. Meta has been trying to close the gap with companies including OpenAI, Anthropic and Google, which have moved quickly with increasingly powerful AI models and agents. The company has already integrated Muse into its overall AI approach. It has its own AI assistant that can search online, make purchases, make travel arrangements, send emails, make payments, and more, and Meta has been working to bring AI into its platforms in deeper ways. The new Wonder models are likely to bolster those efforts as competition becomes more focused on systems that can reason, perform more complex tasks and act on behalf of users. Meta's stock jumped nearly 4% after the announcement in afternoon trading. It also coincides with a period when competitors are aggressively adding more models to their offerings. Last week, Anthropic announced Claude Haiku 5.5, and Google recently announced a new agent called Gemini for workplace tasks.