Meta has introduced Muse Spark 1.1, its newest multimodal reasoning model, positioning the release as a major step in its push toward monetizing artificial intelligence. Announced on July 9, 2026, the model is designed to handle agentic tasks with improved automation, tool use, coding, and cross-session context retention, while requiring less manual guidance to operate unfamiliar interfaces.
A Push Toward Business Automation
Meta says Muse Spark 1.1 can intelligently decide when to write scripts versus interact directly with an interface, allowing it to complete multi-step digital tasks more efficiently. The company frames this as part of a broader ambition to build “personal superintelligence” tools capable of planning and coordinating actions across external apps on a user’s behalf. Meta VP of AI products Vishal Shah highlighted the model’s strength in orchestrating complex personal and business workflows, an area Meta is eager to expand given its enormous AI spending commitments.
Monetization Pressure Builds
Notably, Muse Spark 1.1 will be Meta’s first AI model to include a paid developer tier, with CEO Mark Zuckerberg promising “aggressive and attractive” pricing. The move comes as Meta faces mounting pressure to justify AI investments that could exceed $1 trillion, even though the company’s revenue remains almost entirely ad-driven. Critics note that Meta’s own agentic tools have stumbled recently — including an incident where AI-powered support agents mistakenly gave hackers access to roughly 20,000 Instagram accounts. Despite such setbacks, Meta plans to lean further into automation, aiming to replace the majority of its content moderation staff with AI by the end of 2026, a strategy that raises fresh questions about reliability as agentic AI takes on greater responsibility.