Meta has introduced its new AI model, Muse Spark, marking a strategic initiative to establish a foothold in the competitive AI landscape, dominated by players like OpenAI and Google. This launch follows the recruitment of Alexandr Wang, former CEO of Scale AI, who now leads Meta Superintelligence Labs, the division responsible for Muse Spark's development.
The model, which was codenamed Avocado during its creation, is designed to prioritize efficiency and performance across various tasks, including science, math, and health-related queries. Meta emphasizes that Muse Spark is not intended to be the most advanced AI model but rather a nimble solution capable of delivering competitive results. The company has invested heavily in AI infrastructure, forecasting capital expenditures between $115 billion and $135 billion for 2026, nearly doubling last year's spending.
Muse Spark will support a range of functionalities, including a digital assistant for Meta's standalone AI application and integration into platforms like Facebook, Instagram, and WhatsApp. It will feature multiple modes to cater to different user needs, from simple inquiries to complex tasks requiring in-depth analysis. Notably, a "Contemplating mode" is set to be introduced, which employs a team of AI agents for tackling more intricate queries.
In addition to enhancing user experience within its applications, Meta plans to monetize Muse Spark through API access for third-party developers, initially offering this to select partners with plans for wider availability in the future. This move aligns with Meta's broader strategy to leverage AI advancements to boost its advertising business and remain competitive in a rapidly evolving market.