Meta is accelerating efforts to launch its next-generation Llama AI model by the end of the year, according to people familiar with the company’s plans. The project, known internally as Llama 4.X or Llama 4.5, is being developed by a team within TBD, one of four units under Meta Superintelligence Labs (MSL), the company’s newly formed AI division focused on advancing toward superintelligence. The goal is to deliver a production-ready version of the model before the year ends. This push follows a lukewarm reception to Meta’s April release of Llama 4, which included the Scout and Maverick models. Some developers criticized the update for underperforming in key areas such as coding, logical reasoning, and instruction-following, prompting internal teams to work on fixing flaws and improving performance. In addition to Llama 4.X, Meta had been developing a more ambitious model called Behemoth as part of the Llama 4 series. However, reports from The Wall Street Journal in May indicated that the rollout of Behemoth was delayed, suggesting the company is prioritizing refinement over rapid release. Meta’s broader AI strategy is centered on MSL, which was announced by CEO Mark Zuckerberg in a June internal memo. By August, the company had restructured its AI teams around four core areas: training, research, product, and infrastructure. Alexandr Wang, who joined Meta from Scale AI and now leads MSL, outlined in a follow-up memo that the TBD team would focus on training and scaling large models to achieve superintelligence, including work on an “omni model”—though details remain scarce. Zuckerberg confirmed progress during Meta’s second-quarter earnings call in July, stating the company is advancing Llama 4.1 and 4.2 while simultaneously building next-generation models expected to push the boundaries of AI in the coming year. The initiative comes amid a high-stakes AI talent war. Meta has reportedly offered multimillion-dollar compensation packages to lure top researchers from rivals like OpenAI and Google DeepMind. Despite the aggressive hiring, the MSL unit has already seen early attrition, with at least eight employees—including researchers, engineers, and a senior product leader—departing in the past two months.