Product Launches
Meta Launches Muse Spark as AI App Surges to Top 5
Meta debuted Muse Spark, its first proprietary large language model built by Alexandr Wang's Superintelligence Labs, sending the Meta AI app from #57 to #5 on the App Store as downloads surged 87% in a single day.
Meta Debuts Its Most Ambitious AI Model Yet
Meta officially launched Muse Spark on April 8, 2026, the first proprietary large language model produced by Meta Superintelligence Labs under the leadership of chief AI officer Alexandr Wang. The model represents a ground-up overhaul of Meta's AI capabilities and a decisive shift from the company's previous open-source Llama strategy toward a proprietary, closed-model approach.
Muse Spark — internally code-named Avocado — was built over nine months by a team Wang assembled after Meta acquired a 49% nonvoting stake in Scale AI for $14.3 billion in June 2025. According to benchmarks published by Meta, Muse Spark is competitive with leading models from OpenAI, Anthropic, and Google across multimodal perception, reasoning, health, and agentic tasks.
Meta AI App Rockets to #5 on the App Store
The impact on consumer adoption was immediate. According to market intelligence provider Appfigures, the Meta AI app surged from #57 to #5 on the U.S. App Store within 24 hours of the Muse Spark launch. Daily U.S. iOS downloads hit approximately 46,000 on April 8, an 87% increase over the previous day.
Meta AI has now been downloaded 60.5 million times worldwide across the App Store and Google Play, with 25 million of those installs occurring in 2026 alone. The app's rapid climb positions it as a serious challenger to ChatGPT and Google's Gemini in the consumer AI assistant race.
"Muse Spark is about building AI that truly understands you — your health questions, your creative projects, your daily routines. We built this from scratch to be personal." — Alexandr Wang, Meta Chief AI Officer
From Open Source to Closed: Meta's AI Pivot
Perhaps the most significant aspect of the launch is Meta's decision to keep Muse Spark proprietary. After years of championing open-source AI through its Llama model family, Meta is withholding Muse Spark's code and weights. The company has stated there is "hope to open-source future versions," but for now, the model remains closed — a strategic reversal that reflects the enormous investment Meta has made in AI capabilities.
Meta's 2026 AI capital expenditures are projected at $115 to $135 billion, nearly double 2025 levels. The company is betting that proprietary models like Muse Spark, deployed across Facebook, Instagram, WhatsApp, Messenger, and Ray-Ban Meta AI glasses, will generate returns that justify the unprecedented spending.
What This Means for Engineers and Job Seekers
Meta's pivot to proprietary models signals a new phase in the AI talent war. Engineers with experience in multimodal AI, on-device inference, and personalized model fine-tuning will be especially sought after as Meta races to differentiate Muse Spark across its product ecosystem. For job seekers, the sheer scale of Meta's AI investment — over $100 billion in capex alone — means thousands of roles across research, infrastructure, and product development remain open.