Meta Introduces Muse Spark, Its First Multimodal Reasoning Model
Meta Superintelligence Labs unveiled Muse Spark, a multimodal reasoning model with tool-use and multi-agent orchestration capabilities.
Meta Superintelligence Labs unveiled Muse Spark, a natively multimodal reasoning model designed for tool-use, visual chain of thought, and multi-agent orchestration. The company describes it as the first step on its scaling ladder and the product of a ground-up overhaul of its AI architecture.
The release signals Meta's ambitions beyond large language models into what it calls "personal superintelligence" — systems that combine reasoning across text and vision with the ability to coordinate multiple agents and invoke external tools. Multi-agent orchestration and native tool-use place Muse Spark in direct competition with similar capabilities from Anthropic, Google, and OpenAI, while the visual chain-of-thought feature suggests Meta is betting on tighter integration between perception and reasoning as a path to more capable AI systems.