Meta's 1M-token Muse Spark Open hits Mac Studio with 42B active params

Meta AI Research released Muse Spark Open today, making the large language model available for immediate use on specific hardware. According to Techmeme, this release brings the model to Apple M5 Ultra Mac Studios for the first time.

The model contains 816 billion total parameters, though only 42 billion are active during operation. Meta AI Research describes it as the most capable US open-weight model. It supports a context window of one million tokens.

Technical specifications

The architecture utilizes 42 billion active parameters out of the 816 billion total. The one-million-token context window allows for extensive text input. The model is available today on Apple M5 Ultra Mac Studios.

Affected organizations

Users with Apple M5 Ultra Mac Studios can access this model. The release brings leading agentic capabilities to desktops for the first time on this hardware, according to Meta AI Research. Organizations using this specific Apple hardware can now utilize the model.

Defender's checklist

Verify that your devices meet the hardware requirements before attempting deployment. Ensure your operating system is updated to support the M5 Ultra architecture if you plan to test the model. Review your data handling policies to confirm that local processing aligns with your security standards for sensitive information.

Operational impact

Teams should evaluate whether local model deployment aligns with current infrastructure capabilities. Determine if existing hardware can support the computational load of active parameters without affecting other business processes. Check whether vendor support agreements cover the use of open-weight models in production environments. Discuss with legal counsel how data privacy policies apply when processing sensitive documents on local devices rather than through managed cloud services.

What to do and how to stay safe: Meta

• Confirm your organization has access to Apple M5 Ultra Mac Studio hardware before planning deployment.
• Review internal policies regarding the use of open-weight models for handling proprietary or sensitive data.
• Test the model's performance with sample datasets to ensure it meets your specific operational needs.
• Monitor official channels from Meta for any future updates or security patches related to the model.

Step-by-step guide: Implementing AI Usage Policies at Work: A Step-by-Step Framework

General security guidance from the InfoSec Ledger newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

What is the parameter count of the new Muse Spark Open model?

The model has 816 billion total parameters and 42 billion active parameters, according to Meta AI Research.

Which hardware supports the initial release of this model?

Muse Spark Open is available on Apple M5 Ultra Mac Studios, as reported by Techmeme.

How large is the context window for this model?

The model supports a context window of one million tokens, enabling large-scale text processing.