Meta introduces advanced AI models to challenge OpenAI and anthropic

On October 12, 2026, Meta launched two new AI models, Muse Wonder and Muse Wonder Full, positioning them as state-of-the-art tools for personal and professional tasks. These models are part of Meta's broader effort to close the gap with leading AI firms like OpenAI, Anthropic, and Google. According to Alexandr Wang, head of Meta's Superintelligence Labs, the models are competitive with the best in the world, particularly in agentic work. Meta emphasized two key features: Active Memory and Active Focus, designed to enhance AI performance over extended periods. The company also open-sourced the weights for Muse Glitter and the technology for deploying Muse agents on hardware. Privacy and security concerns remain, as agents that remember personal information pose new challenges. The success of these models will depend on their performance in independent tests and their adoption by Meta's vast user base. Meta's long-term goal, as outlined by CEO Mark Zuckerberg, is to develop 'personal superintelligence'—AI that understands individual goals and preferences. While Meta has not disclosed plans to open-source the Wonder models, it has already taken steps to share technology with developers. The company's significant investment in AI research and infrastructure is being tested by these new models, which represent a critical milestone in its competitive strategy against industry leaders.