Meta has officially introduced Muse Spark, the first AI model from its newly formed Meta Superintelligence Labs (MSL), the division led by Alexandr Wang following Meta's multibillion-dollar Scale AI deal. Muse Spark is a natively multimodal reasoning model capable of handling text, voice, and image inputs, with support for tool use, visual reasoning, and multi-agent orchestration for more complex tasks. According to Meta, the model now powers the upgraded Meta AI assistant across meta.ai and the Meta AI app, with rollout planned across WhatsApp, Instagram, Facebook, Messenger, and Meta's AI glasses in the coming weeks. The company says Muse Spark was built after rebuilding its AI stack from the ground up over the past nine months. Benchmark disclosures suggest Muse Spark is competitive with leading frontier models in reasoning, multimodal understanding, and health-related tasks, though Meta acknowledges it still trails top competitors in coding and some agentic benchmarks. Notably, the company emphasized the model's strength in health reasoning, developed in collaboration with physicians. Unlike the open-weight Llama family, Muse Spark is currently proprietary, though Meta says future open-source variants may follow. Why it matters: Muse Spark marks Meta's clearest sign yet that it is re-entering the frontier AI race with a new architecture and strategy beyond Llama. While not yet the outright market leader, the launch gives Meta a stronger foundation for embedding advanced AI across its massive consumer ecosystem and signals that Zuckerberg's AI reset may be starting to show results.