In early April, Meta unveiled the Muse Spark, its latest AI model and the first major release under AI Chief Officer Alexandr Wang. The model performs competitively in other benchmarks but still trails OpenAI’s GPT-5.4 Pro and Google’s Gemini 3.1 Pro, raising questions about whether Meta is lagging behind. Wang believes framing misses the point.
“The new Muse Spark model that we released is not in the category of flagship models,” Wang said during an interview on stage at the Bloomberg Tech Summit in San Francisco yesterday (June 4). “But we believe it’s a very exciting data point in the trajectory, and we expect the future models we release to compete with the best models in the world.”
He described Muse Spark as “an appetizer.” Asked when the entrant would arrive, Wang replied: “We’re cooking it up. We’re seeing exciting and promising results from the training program right now.”
Muse Spark marks a Meta shift. It is the company’s first model and is used only in Meta products, rather than being released openly as previous systems were. The model is designed to handle text, images, video and audio, and to support complex, multi-step operations, including shopping features tied to Instagram and Facebook content.
The release follows a serious stretch of Meta’s AI efforts. Llama 4, launched in April 2025, was widely criticized. Two months later, Mark Zuckerberg hired Wang to lead his newly formed Superintelligence Labs and reset its strategy.
Wang said the group is focused on data augmentation: increasing data, computing power and research to improve development. And Muse Spark sits early in that process.
The barrier at the border is not money, Wang said. “It’s about continuing to scale data, computing … and continuing to scale with research. All the labs are dramatically scaling up their models, and we’re on a very fast track because we’ve been doing all this work for the past year.”
Meta supports that approach by spending a lot of money. The company expects spending of $125 billion to $145 billion in 2026, up from $72.2 billion in 2025, and is targeting more than 1.3 million GPUs and nearly one gigawatt of AI computing capacity.
The switch to a closed model also reflects security concerns. During development, the Muse Spark implemented internal warnings, including potential biological hazards.
“When a company launches a model into production, we have many ways to mitigate some of these risks,” he said. “It’s very difficult to do that when you open source the model.”
Meta has not completely abandoned open source AI and continues to develop models it considers safe to release. Whether its Llama breed will continue is unknown. “We’re having interesting debates about internal marketing,” Wang said, “and nothing to share right now.”
Wang said Muse Spark’s strengths lie in the power of multitasking, health-related applications and creative coding—such as generating simple games or digital tools. Those areas support Meta’s broader push for AI agents.
The company is “doubling down” on agents, aiming to create what Wang calls “the best agents for everyone around the world.” He said he uses such tools to manage his life and communicate with friends.
The push comes with major interior changes. In May, Meta notified about 8,000 employees of layoffs and assigned about 7,000 more to AI-focused roles as part of a broader restructuring.
“It’s incredibly difficult to say goodbye to my teammates,” Wang said. “We take nothing for granted.”
