Meta and Apple Collaborate on Muse Spark Open AI Model
Tech rivals partner to optimize a powerful open-weights model for Apple hardware.
1Strategic Tech Partnership
Meta and Apple have formed an unexpected alliance to launch Muse Spark Open, an open-weights AI model. This collaboration bridges Meta's software development with Apple's hardware expertise. By working together, the companies aim to ensure the model runs efficiently on high-end Apple silicon, specifically the M5 Mac Studio Ultra hardware.
2Model Technical Specifications
Muse Spark Open features 42 billion active parameters and a massive context window of one million tokens. Designed for multimodal tasks, the model excels in agentic and coding workflows. Developers can access the model weights on HuggingFace under Meta's specific license, allowing for deep customization and integration into various technical environments.
3Hardware Requirements and Costs
Running this model locally requires significant computational power, specifically two M5 Mac Studio Ultra units. With each unit priced at $5,499, the total hardware investment for a functional setup reaches nearly $11,000. This high barrier to entry highlights the intensive resource demands required to execute large-scale, open-weights AI models on-device.
4Divergent Corporate Philosophies
The partnership is notable given the companies' historically conflicting paths regarding data privacy and business models. Apple has long prioritized user privacy, famously costing Meta $10 billion in 2022 through app tracking transparency features. Despite these past tensions, both firms are now aligning their distinct strengths to maintain relevance in the AI race.
5Focus on On-Device AI
The collaboration centers on the growing importance of on-device AI, which operates without relying on external cloud services. This approach allows businesses to maintain greater control over sensitive data while potentially reducing long-term subscription costs. By keeping processing local, developers can build more secure systems tailored to their specific operational needs.
6Apple's Hardware-First Strategy
Apple is positioning itself as the primary hardware provider for the AI era rather than competing solely on software tools. Following the popularity of Mac Minis for AI tasks in early 2026, the company has consistently optimized its chip architecture. The M5 series demonstrates Apple's commitment to supporting compute-intensive AI workloads.
7Industry Leadership Perspectives
Meta's chief AI officer Alexandr Wang described the collaboration as a positive shift in industry relations. Meanwhile, Apple CEO John Ternus emphasized the technical necessity of the partnership for advancing on-device capabilities. This diplomatic alignment suggests a mutual recognition that hardware and software must be tightly integrated to achieve high-performance AI.
8Future Market Implications
This partnership signals a shift where Meta focuses on open-weights model distribution while Apple secures its role as the essential hardware foundation. By enabling complex AI agents to run locally, this move could disrupt traditional cloud-dependent AI workflows. The industry will likely watch how these two giants navigate their ongoing competitive tensions.
Original article: CNET — Meta and Apple Collaborate on Muse Spark Open AI Model