Meta Debuts Muse Spark 1.1 With Paid AI Tier

Meta has unveiled Muse Spark 1.1, its latest multimodal reasoning model, marking a fresh step in the company’s push to expand and eventually profit from its artificial intelligence efforts. Announced on July 9, 2026, the model is designed for agentic tasks and brings notable upgrades in tool use, computer navigation, coding, and multimodal understanding.

Built for Automation and Agentic Tasks

According to Meta, the model can retain context across sessions, adjust to changing requirements, and operate unfamiliar interfaces with little human guidance. The company said it trained Muse Spark 1.1 to decide when to write automated scripts versus interact directly with an interface, aiming for greater efficiency in multi-step digital tasks. Meta framed the release, alongside its recent Muse Image launch, as part of a broader ambition toward building personally useful AI systems.

A Push Toward Monetization

Muse Spark 1.1 will be Meta’s first AI model to offer a paid developer tier, a notable shift for a company where the vast majority of revenue still comes from advertising. CEO Mark Zuckerberg has indicated pricing will be aggressive to attract subscribers. The move comes as Meta faces mounting pressure to justify enormous AI spending, with some estimates suggesting it could take over a decade to recoup costs even under optimistic revenue projections.

The launch also arrives amid scrutiny of Meta’s agentic AI tools following a security lapse last month, when automated support agents on Instagram mistakenly gave hackers access to roughly 20,000 accounts. Despite such incidents, Meta continues to expand its reliance on AI, including plans to automate a large share of its content moderation work by the end of 2026.