Meta unveiled its latest artificial intelligence model on Wednesday, its first since assembling a costly team to help the company better compete with its AI rivals.
The new model, known as Muse Spark, was created by Meta Superintelligence Labs, a team of expensive AI leaders and developers assembled last year from other AI companies and led by Alexandr Wang, the co-founder and CEO of Scale AI. Wang joined the Meta AI team after Meta agreed to make a $14.3 billion investment in Scale AI.
The model, known internally as Avacodo, now powers the Meta AI app and the Meta AI website and will soon power Meta apps WhatsApp, Instagram, Facebook, Messenger and AI glasses. Meta called the new model “an early data point on our trajectory,” adding that it has larger models in development.
“This initial model is small and fast by design, yet capable enough to reason through complex questions in science, math, and health,” Meta said in its announcement. “It is a powerful foundation, and the next generation is already in development.”
Meta has been investing increasingly large sums to compete in the ever-evolving AI landscape, creating a “superintelligence team” that works toward achieving artificial general intelligence, or AGI. The release of Muse Spark comes after the company’s disappointing release of the Llama models last year.
Meta has bet big on AI to catch up and keep pace with other AI builders. Google leapfrogged its rivals in November with its Gemini 3 model, showcasing its impressive coding and research abilities. OpenAI was quick to follow with updates to GPT-5.
