admin In an exclusive development, Meta Platforms’ ambitious artificial intelligence division, Superintelligence Labs, has delivered its first high-profile AI models for internal use this month. Chief Technology Officer Andrew Bosworth announced this significant milestone, revealing the progress during a press briefing at the World Economic Forum in Davos.
Bosworth expressed considerable optimism about the early output from the team, which was formed last year as part of a major strategic push by CEO Mark Zuckerberg to accelerate the company’s AI capabilities.
The CTO noted that the team, having been at work for barely six months, has already produced models that show tremendous promise. This internal delivery marks a crucial first step for the lab, which was established after Meta faced criticism over the performance of its previous Llama 4 model, a period during which rivals like Google seized momentum in the competitive AI landscape. The early results are seen as a positive initial return on the company’s aggressive 2025 bets, which involved a leadership overhaul and a high-stakes talent acquisition drive.
While Bosworth did not specify which models were delivered, recent reports have indicated Meta is developing a text-based model codenamed Avocado and a multimedia model focused on images and video known as Mango.
Despite the encouraging progress, Bosworth cautioned that the technology is not yet complete, emphasizing that a substantial amount of post-training work remains to refine the models for practical use, both internally and for future consumer applications. This milestone nonetheless signals that Meta’s reorganized AI efforts are beginning to yield tangible results as the company vies for leadership in the transformative and lucrative field of artificial intelligence.
Magazine 
