Meta has introduced Muse Spark, its first model from Meta Superintelligence Labs, established under Alexandr Wang after acquiring a stake in Scale AI for $14.3 billion. This model, developed over nine months, is natively multimodal and features a “Contemplating” reasoning mode. Unlike Meta’s previous open-source Llama models, Muse Spark is closed source.
The launch of Muse Spark follows the creation of Meta Superintelligence Labs in June 2025, where Wang was appointed as the first chief AI officer. The mission was to catch up with AI leaders like OpenAI and Google. In his announcement, Wang emphasized that Muse Spark represents an entirely new AI stack, rebuilt from the ground up.
Previously known internally as “Avocado,” the model had been delayed but is now released, suggesting competitive readiness. Wang describes Muse Spark as the first in a series of models, emphasizing process over final product.
Muse Spark is multimodal, accepting voice, text, and image inputs, and operates in both a fast query mode and a “Contemplating” mode that employs multiple sub-agents for parallel reasoning. With ten times greater efficiency than Llama 4 Maverick, it uses a method called “thought compression.”
Benchmark results show Muse Spark ranking fourth on the Artificial Analysis Intelligence Index v4.0. It performs best on CharXiv Reasoning and HealthBench Hard, aligning with Meta’s strategic advantage in data and health collaboration.
A focused shopping mode utilizes Meta’s user data for personalized recommendations, while health capabilities leverage partnerships with physicians for precise analyses.
Muse Spark is currently powering the Meta AI app and website, with future deployment across platforms like Facebook, Instagram, and WhatsApp.
Meta’s decision to keep Muse Spark closed source is notable, deviating from the open-source tradition of the Llama series. The company hints at possible future open-source releases but recognizes the competitive edge of proprietary innovation.
The $14.3 billion investment in Scale AI and continued capital spending support this development. Despite a 9% rise in Meta shares, driven by broader market trends, the focus on health and visual understanding demonstrates where Meta leads. This strategic choice highlights Meta’s targeted approach to AI competitiveness.