Meta Enters the Frontier AI Race With Muse Wonder
Meta has unveiled Muse Wonder and Muse Wonder Full, its first frontier-scale AI models and the company's most ambitious challenge yet to OpenAI, Anthropic, and other leading AI labs. Developed by Meta's Superintelligence Lab, the models are designed to deliver stronger performance on complex tasks such as coding while matching top-tier competitors on key industry benchmarks. Meta claims the new models achieve results comparable to systems like GPT-6 Astra and Fable 5.1, marking a significant step up from the smaller Muse Spark model released earlier. A major focus of the launch is "active memory," which aims to help AI assistants retain and synthesize user information over extended periods rather than treating every interaction as isolated. According to Meta, Muse Wonder outperforms rival models on its preferred MemoryBench evaluation designed to measure long-term recall and knowledge retention. The models will also serve as the foundation for Meta's Muse AI agent, reflecting the company's growing emphasis on persistent digital assistants. Access to Muse Wonder will be provided through Meta's API, while the more advanced Wonder Full variant will be reserved for Muse Code subscribers.
Read the full story on TechCrunch →