Meta Collaborates with Apple on Open-Source AI Model

Meta has made a surprising move in the world of artificial intelligence by collaborating with Apple's hardware team to create an open-source AI model. The new model, called Muse Spark Open, is owned by Meta and has been designed to run efficiently on Apple's M5 Mac Studio Ultra.

Muse Spark Open is an open-weights AI model, which allows developers to understand its inner workings and customize it to suit their needs. This level of transparency is a significant departure from the usual approach taken by tech companies in developing AI models. The model boasts 42 billion active parameters and can handle tasks involving multimodal capabilities.

These multimodal capabilities enable the model to excel at agentic and coding tasks, making it an attractive tool for engineers and developers. Meta claims that its open-weights design allows for a high degree of flexibility, allowing users to adapt the model to their specific needs. The model's weights are now publicly available on HuggingFace under Meta's Muse Model license.

To run Muse Spark Open on Apple computers, users will need two Mac Studios with M5 Ultra chips. These devices come at a hefty price, with the base cost for one device starting at $5,499. This is due to the high-end specifications of the devices, which include 96 GB of memory and 1 TB of storage.

The requirement for two devices to run the model may be a significant barrier for some users, particularly those who are not looking to invest in expensive hardware. Furthermore, there is no information available yet on whether Apple will offer devices with the M5 Ultra chip that have a full 512 GB capacity, which would likely increase the cost even further.

Meta's shift towards open-weights AI models has been well-documented, but its collaboration with Apple on this new project marks a significant departure from their typically divergent approaches to artificial intelligence development.

While Meta has been aggressively investing in AI research and development, pouring billions of dollars into the field, Apple has taken a more measured approach, prioritizing user privacy over AI-driven initiatives. This dichotomy is particularly evident in their respective business models, with Meta's reliance on targeted advertising starkly contrasting with Apple's emphasis on data protection.

The partnership between these two tech giants appears to be driven by a shared interest in on-device AI, which enables the running of AI models and tools directly on local hardware rather than relying on cloud services. This approach has several advantages, including greater control over data security and potentially lower costs for large-scale operations.

On-device AI requires some setup and configuration from developers and businesses, but it offers a more secure and cost-effective alternative to cloud-based solutions. The collaboration between Meta and Apple is likely to accelerate the development of on-device AI capabilities, which could have significant implications for various industries that rely heavily on AI-driven applications.

The partnership's focus on on-device AI suggests that both companies are committed to advancing this technology in tandem with their respective goals. As they work together to develop more sophisticated on-device AI models and hardware, it will be interesting to see how this collaboration impacts the broader AI landscape and the industries that rely on its innovations.

The partnership between Meta and Apple to develop a new open-source AI model has been met with surprise by some, but it's not entirely unexpected given Apple's focus on hardware that can support AI tasks. The company's Mac Minis have been popular among AI enthusiasts since the beginning of 2026, when the OpenClaw agent was being widely used.

Apple's emphasis on its chip technology and its ability to power impressive performance has been demonstrated in various products, including the M5 MacBook Air released in March. The Mac Studio, launched in August, is another example of Apple's hardware capable of running compute-intensive AI tasks, further solidifying its position in the AI landscape. This collaboration will likely have a significant impact on the broader AI industry and its ongoing innovations.