Meta has introduced a new artificial intelligence model called Muse Spark, describing it as the first step in a broader overhaul of its AI efforts.
The model is now available on the web and through the Meta AI app and is expected to improve over time with additional features.
Built by Meta Superintelligence Labs
Muse Spark is the first model developed by Meta Superintelligence Labs, a division created last year after CEO Mark Zuckerberg expressed concerns over the company’s progress in AI, particularly its Llama models.
To strengthen its position, Meta recruited Alexandr Wang to lead the division and invested $14.3 billion in Scale AI for a 49% stake.
Multi-Agent System
Muse Spark introduces a multi-agent system in which multiple AI agents collaborate on the same task.
Meta plans to add a “Contemplating” mode, designed to handle more complex problems by allowing extended reasoning without significantly increasing response time.
The system achieves this by running parallel agents that collaborate to generate faster and more detailed outputs.
Competitive Positioning
Unlike some competitors that place advanced AI features behind a paywall, Meta has not yet clarified whether Muse Spark will follow a similar pricing strategy.
The launch comes as Meta attempts to compete more directly with offerings from companies such as OpenAI and Anthropic.
Health Use Cases
Meta said Muse Spark could be used to assist with health-related questions, aligning with a broader industry trend among AI developers.
The model is also designed to perform well in visual STEM-related tasks, enabling use cases such as creating simple interactive experiences or troubleshooting household devices.
Privacy Issues
Access to Muse Spark requires users to log in with an existing Meta account, such as Facebook or Instagram.
While the company has not explicitly stated that personal data from these accounts will be used, Meta has historically trained its systems on public user data and has positioned Muse Spark as a personal AI product.
In addition to its investment in Scale AI and leadership changes, Meta has recruited researchers from companies including OpenAI, Anthropic, and Google.
Zuckerberg stated that plans include releasing more advanced models, including open-source options, and building AI systems that can act as agents capable of completing tasks rather than only answering questions.
