Meta's new chief AI officer considers scrapping its most powerful open-source AI model, Behemoth—NYT.

Meta's new chief AI officer considers scrapping its most powerful open-source AI model, Behemoth—NYT.

Meta's new chief AI officer, is reportedly considering scrapping the company's most powerful open-source AI model, Behemoth, according to a recent New York Times article. The move comes as Meta ramps up its efforts to regain a leadership position in the AI space, following recent setbacks in model performance compared to rivals like OpenAI and Google DeepMind. CEO Mark Zuckerberg has been personally leading Meta's charge in AI, identifying talent and compute as key areas for improvement. The company has been aggressively hiring top AI talent, offering compensation packages worth over £750,000 and even exceeding £1 million in some cases [1]. This includes generous base salaries, stock awards, and performance bonuses, making these offers among the most lucrative in the tech industry. Meta's open-source strategy has gained attention, with the company releasing its LLaMA models to the research and developer community. However, critics have questioned whether open-sourcing highly capable models could pose security and misuse risks. Despite these concerns, Meta's approach has resonated with some in the AI field, positioning the company as a champion of transparency and open innovation. Zuckerberg's consideration to scrap Behemoth is part of a broader strategy to reinvent Meta's AI capabilities. The company has been investing heavily in compute infrastructure, reportedly on track to deploy 600,000 GPUs by the end of 2025, giving it one of the largest AI training capabilities in the world [1]. This move aims to both train cutting-edge models and attract engineers who want access to the best tools and resources. The fierce competition in the AI space has sparked concern about rising salaries and a potential brain drain from academia and smaller startups. Meta has been actively recruiting from top universities and rival labs, sometimes offering compensation that far exceeds what researchers could earn in the public sector. However, this aggressive hiring strategy has also led to criticism, with some accusing Meta of poaching talent from competitors like OpenAI. Meta's new AI strategy, including the potential scrapping of Behemoth, is seen as a significant shift in the company's approach to AI. While the move is still under consideration, it reflects Meta's determination to regain its edge in the fast-evolving race to dominate next-generation technology. The company's future success in AI will depend on its ability to balance innovation, transparency, and ethical considerations in its AI models and strategies. References: [1] Saf Malik. "Explained: Why Meta is offering huge pay packages to AI talent in the race to catch rivals." Capacity Media. July 14, 2025. https://www.capacitymedia.com/article-explained-meta