After months of repeated attempts, Meta has finally unveiled a new artificial intelligence model—Muse Spark—that can compete with its rivals. This model not only represents a technological advancement but also marks a significant shift in the company's strategic direction, signaling the end of an era.
Muse Spark Becomes Meta's New AI Flagship
The project is led by 28-year-old Alexandr Wang, founder of Scale AI and Meta's new head of AI. After acquiring a 49% stake in Scale AI for approximately $14 billion, Wang joined the company and is now responsible for Superintelligence Labs, a department dedicated to pushing the limits of Meta's development in artificial intelligence.

In terms of performance, Meta claims that Muse Spark can accomplish the same tasks as its predecessor, Llama 4 Maverick, but with significantly reduced computational resource consumption. Internal testing has shown that Muse Spark has already reached or even surpassed similar solutions from Google, OpenAI, and Anthropic in certain metrics. Independent evaluations indicate that the model performs exceptionally well in the medical field, although it still has shortcomings in programming capabilities.
Unspoken Strategic Shift
This release is far from a simple technical update; Muse Spark breaks Meta's long-standing commitment to open-source traditions. Unlike the Llama series, Muse Spark adopts a closed and proprietary model. Some partners can preview access via API, but Meta has not confirmed whether it will be publicly released. The company mentioned the possibility of launching an open-source version in the future, but no specific timeline has been provided.

This shift reflects the immense pressure Meta faces in the AI race. The Llama 4, released in 2025, underperformed expectations, and the costly Behemoth project, with parameters reaching 200 billion, has yet to see the light of day. In the face of setbacks, management has decided to accelerate its pace while safeguarding the company's core technological assets.
To support this strategy, Meta plans to invest $135 billion in AI research and development by 2026, nearly double the amount from the previous year. Such a massive investment also brings societal implications. The Reality Labs department, previously responsible for the metaverse, has already spent $80 billion and undergone hundreds of layoffs. Multiple sources indicate that there may be plans for over 20% more layoffs in the future, although Meta has denied this, leaving lingering doubts in the public eye.
Meta executives warn that Muse Spark is just the beginning, with more advanced models set to be released throughout the year, as the AI race accelerates once again.

