Meta’s Superintelligence Lab is reportedly reevaluating its position on open source AI, according to a report by The New York Times.

Discussions within the lab have raised the possibility of moving away from the company’s open source AI model, Behemoth, in favor of developing closed models.

While Meta has completed training on Behemoth, sources indicate that its release has been delayed due to unsatisfactory internal performance.

Testing on the model has reportedly been paused since the launch of the Superintelligence Lab.

However, any decision to shift towards closed models would require approval from CEO Mark Zuckerberg.

A Meta spokesperson said that the company’s commitment to open source AI remains unchanged.

They added that Meta intends to maintain a mix of open and closed models in its development pipeline. No comment was provided regarding the potential shift away from Behemoth specifically.

Meta has not confirmed any changes to its AI strategy at this time.

🔗 Source: TechCrunch

🧠 Food for thought

1️⃣ Meta’s pivot mirrors broader AI industry tension between openness and commercialization

Meta’s potential shift from open-source models like Llama to closed systems reflects an industry-wide tension between collaborative innovation and commercial pressures.

The financial gap is significant, as since 2020, closed-source AI developers have secured $37.5 billion in venture funding compared to just $14.9 billion for open-source developers, creating strong incentives for companies to protect their intellectual property 1.

This reflects a broader pattern where companies initially embrace openness for community goodwill and development acceleration, then shift toward closed models once commercial viability becomes the priority.

FAIR (Facebook AI Research) built its reputation on open collaboration since its founding, with tools like PyTorch becoming industry standards and helping establish Meta’s AI credibility 2.

However, as competition intensifies and the costs of developing frontier AI models continue to escalate into the billions, the business case for keeping advanced capabilities proprietary grows stronger.