Meta's new Muse Spark model drove an 87% iOS download spike and 450%+ web traffic surge, signaling genuine consumer traction.
Meta launched Muse Spark, its first model under new Superintelligence Labs head Alexandr Wang (poached from Scale AI), on April 8, 2026. The Meta AI app jumped from No. 57 to No. 5 on the U.S. App Store within 24 hours, with ~46,000 U.S. iOS downloads — an 87% day-over-day increase. U.S. daily web visitors rose more than 450% day-over-day and hit an all-time high. Muse Spark supports multimodal input (voice, text, images), multi-agent orchestration, visual coding, and reasoning tasks in health, science, and math.
Muse Spark's multi-agent orchestration and visual coding capabilities are live on web and mobile. If Meta exposes this via API (likely through the Llama ecosystem), it gives developers a free or low-cost path to multimodal agent pipelines — without paying OpenAI or Anthropic rates. The model's positioning around reasoning, health, and science tasks suggests benchmark-grade performance, not just consumer fluff.
Go to meta.ai this week and test Muse Spark on a complex multi-step reasoning task you currently run through GPT-4o or Claude — compare output quality and note latency before deciding whether to wait for an API release.
Go to meta.ai in your browser and sign in with a Meta account
Tags
Sources
Also today
Signals by role
Also today
Tools mentioned