Meta rolls out its first AI Model - Muse Spark to compete with rivals

Meta has introduced its latest artificial intelligence model after Ceo, Mark Zuckerberg spent a multibillion-dollar to transform the company’s AI organization to keep pace with rivals.  

The model named Muse Spark was created by Meta Superintelligence Labs – a newly formed team of AI researchers headed by Chief AI Officer Alexandr Wang. The new model will empower the Meta AI chatbot and it will be a closed model, meaning its design and code would not be made public.     

With this announcement, Meta shares jumped 6% in New York.

The model has been tested by the company’s AI lab known as MSL. Zuckerberg appointed Wang as part of a $14 billion investment into Scale AI last year, following a number of setbacks that left him frustrated that the company was not keeping pace with competitors. He also spent billions on AI talent and committed tens of billions more on infrastructure like data centers to catch up.

As per news reports, a company executive said, Meta has to keep its AI division agile by granting its researchers autonomy and reduced its typical management-heavy organizational structure. Wang has around 100 direct reports, said the executive.   

He conceded that the model was not as efficient as ChatGPT, Claude or Gemini in some areas, but said that it was early in the company’s execution. The model is “an early data point on our trajectory,” with several larger models in development, Meta said in a blog post.   

Muse Spark was built in a time span of nine months and it is seen as a refresh of Meta’s AI strategy by executives. Earlier, Meta focused on its open-source Llama models. Wang is a proponent of closed models, and while Meta still plans to build open-source ones in the future, it is also considering selling API access to Muse Spark, according to the company executive.

Meta will not charge anything from its AI chatbot users but the company may implement subscription fees in the future, the executive said.

The model has been trained using third-party open-source models like Qwen, ChatGPT and Gemini.        

“Like others across the industry, Meta uses techniques like distillation with strict safeguards in place to learn from openly available AI models and improve our own,” said a Meta spokesperson.

Alike competitors, Muse Spark offers various levels of reasoning like “Instant” mode, “Thinking” mode and “Contemplating” mode — the last providing research-grade responses. The Meta executive said the model is good at answering questions about science, health and math, but weaker at coding.

The model will be used to improve Meta’s fleet of apps, from Instagram to Facebook to WhatsApp. It is already powering an early-stage shopping agent that aims to help users better search for products like clothing or furniture. The agent cannot yet make purchases on a user’s behalf, and is still in early stages, the executive said.