2 min read
Meta Superintelligence Lab (MSL), the billion-dollar initiative launched by Facebook-parent Meta, is racing against time to deliver its first product, which could arrive by the end of this year. Reports suggest the debut offering will be a next-generation Llama AI model, marking the first major project under MSL.
According to a report by Business Insider, a team within the Meta Superintelligence Lab, internally known as TBD (to be determined), is developing the Llama 4.X models and preparing them for production by year-end. This timeline has become a top priority for the company as it seeks to showcase progress from its newly created unit.
Some employees are also referring to the new AI model as Llama 4.5, suggesting a significant leap forward from earlier versions. The urgency partly stems from the lukewarm reception of Meta’s previous releases—Scout and Maverick—which launched in April but faced criticism for underperforming in areas like coding, reasoning, and instruction-following. In response, the TBD team is simultaneously addressing bugs in the existing Llama 4 models while advancing the Llama 4.X project.
Meta CEO Mark Zuckerberg highlighted during the company’s second-quarter earnings call in July that MSL is working toward multiple next-generation models. He said progress was being made on Llama 4.1 and 4.2, while in parallel the lab is focusing on innovations that could “push the frontier” in the coming year.
The Meta Superintelligence Lab was established in June, accompanied by a hiring spree that included multimillion-dollar compensation offers to attract top AI talent from competitors such as OpenAI, Apple, and Google DeepMind. However, the initiative has faced early setbacks, with several researchers, engineers, and even a senior product leader departing within weeks of its launch.
Despite these challenges, Meta remains determined to position its Superintelligence Lab at the forefront of the AI race, with the Llama 4.X release expected to be a major test of its strategy.
