With the first major release of Alexandr Wang, the head of Meta's Superintelligence Lab, Muse Spark has officially launched, marking a period of explosive growth for Meta in the AI field. Driven by the release of this model, the ranking of Meta AI apps in app stores has seen a dramatic rise.
Double Surge in Ranking and Downloads
Market intelligence data shows that before the release of Muse Spark, Meta AI was ranked 57th in the US App Store. Within less than 24 hours after the model's release, its ranking rapidly climbed to 5th. According to Sensor Tower estimates, on April 8th, the app's downloads on iOS increased by 87% compared to the previous day, reaching about 46,000 times; daily website traffic increased by more than 450%, setting a new historical high.
Performance Leap: Multimodal and Logical Reasoning
As the successor to Llama4, Muse Spark marks a significant technological advancement by Meta. The model currently supports three input methods: audio, text, and images, and performs well in health consultations, complex scientific and mathematical problem solving. Notably, it also has the ability to assist in visual programming, allowing users to create websites and mini-games with just prompts. In addition, the model introduces a sub-agent collaboration mechanism, capable of launching multiple sub-agents to handle complex tasks.
Strategic Transformation with Heavy Investment
This success is backed by Meta's ambitious strategic initiatives. Last year, Meta hired Alexandr Wang from Scale AI to lead the AI reform and invested 14.3 billion dollars in Scale AI. Although Meta AI still temporarily lags behind ChatGPT, Claude, and Gemini in overall rankings, its global installation count has reached 60.5 million, with India being its largest market.
Meta plans to roll out Muse Spark across its entire hardware and social platforms, including WhatsApp, Instagram, Facebook, and Meta AI glasses, in the coming weeks. With a comprehensive upgrade of the user interface and interaction experience, Meta is trying to redefine AI assistants in the multi-tasking domain through the "mode switching" feature.
